\input texinfo @c -*- mode: texinfo; coding: utf-8 -*- @comment %**start of header @setfilename ../../info/eintr.info @c setfilename emacs-lisp-intro.info @c sethtmlfilename emacs-lisp-intro.html @settitle Programming in Emacs Lisp @include docstyle.texi @syncodeindex vr cp @syncodeindex fn cp @finalout @include emacsver.texi @c ================ How to Print a Book in Various Sizes ================ @c This book can be printed in any of three different sizes. @c Set the following @-commands appropriately. @c 7 by 9.25 inches: @c @smallbook @c @clear largebook @c 8.5 by 11 inches: @c @c smallbook @c @set largebook @c European A4 size paper: @c @c smallbook @c @afourpaper @c @set largebook @c (Note: if you edit the book so as to change the length of the @c table of contents, you may have to change the value of 'pageno' below.) @c <<<< For hard copy printing, this file is now @c set for smallbook, which works for all sizes @c of paper, and with PostScript figures >>>> @set smallbook @ifset smallbook @smallbook @clear largebook @end ifset @c ================ Included Figures ================ @c If you clear this, the figures will be printed as ASCII diagrams @c rather than PostScript/PDF. @c (This is not relevant to Info, since Info only handles ASCII.) @set print-postscript-figures @c clear print-postscript-figures @comment %**end of header @c per rms and peterb, use 10pt fonts for the main text, mostly to @c save on paper cost. @c Do this inside @tex for now, so current makeinfo does not complain. @tex @ifset smallbook @fonttextsize 10 @end ifset \global\hbadness=6666 % don't worry about not-too-underfull boxes @end tex @c For next or subsequent edition: @c create function using with-output-to-temp-buffer @c create a major mode, with keymaps @c run an asynchronous process, like grep or diff @c For 8.5 by 11 inch format: do not use such a small amount of @c whitespace between paragraphs as smallbook format @ifset largebook @tex \global\parskip 6pt plus 1pt @end tex @end ifset @c For all sized formats: print within-book cross @c reference with ``...'' rather than [...] @c This works with the texinfo.tex file, version 2003-05-04.08, @c in the Texinfo version 4.6 of the 2003 Jun 13 distribution. @tex \if \xrefprintnodename \global\def\xrefprintnodename#1{\unskip, ``#1''} \else \global\def\xrefprintnodename#1{ ``#1''} \fi % \global\def\xrefprintnodename#1{, ``#1''} @end tex @c ---------------------------------------------------- @dircategory Emacs lisp @direntry * Emacs Lisp Intro: (eintr). A simple introduction to Emacs Lisp programming. @end direntry @c When printing, define edition-number to be the printed edition @c number, titlepage-edition-number to be the spelled out edition @c number suitable for the title page, and update-date to be the date, @c in the preferred style for these. E.g., run the shell command: @c texi2any -D 'edition-number 3.11' \ @c -D 'titlepage-edition-number Revised Third Edition' \ @c -D 'update-date 31 March 2020' @c This relates mainly to the published book sold by the FSF. @copying This is @cite{An Introduction to Programming in Emacs Lisp}, for people who are not programmers. @sp 1 @ifset edition-number Edition @value{edition-number}, @value{update-date} @end ifset @sp 1 Distributed with Emacs version @value{EMACSVER}. @sp 1 Copyright @copyright{} 1990--1995, 1997, 2001--2020 Free Software Foundation, Inc. @sp 1 @iftex Published by the:@* GNU Press, @hfill @uref{https://www.fsf.org/licensing/gnu-press/}@* a division of the @hfill email: @email{sales@@fsf.org}@* Free Software Foundation, Inc. @hfill Tel: +1 (617) 542-5942@* 51 Franklin Street, Fifth Floor @hfill Fax: +1 (617) 542-2652@* Boston, MA 02110-1301 USA @end iftex @ifnottex Printed copies available from @uref{https://shop.fsf.org/}. Published by: @example GNU Press, https://www.fsf.org/licensing/gnu-press/ a division of the email: sales@@fsf.org Free Software Foundation, Inc. Tel: +1 (617) 542-5942 51 Franklin Street, Fifth Floor Fax: +1 (617) 542-2652 Boston, MA 02110-1301 USA @end example @end ifnottex @sp 1 ISBN 1-882114-43-4 @quotation Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.3 or any later version published by the Free Software Foundation; there being no Invariant Section, with the Front-Cover Texts being ``A GNU Manual'', and with the Back-Cover Texts as in (a) below. A copy of the license is included in the section entitled ``GNU Free Documentation License''. (a) The FSF's Back-Cover Text is: ``You have the freedom to copy and modify this GNU manual. Buying copies from the FSF supports it in developing GNU and promoting software freedom.'' @end quotation @end copying @c half title; two lines here, so do not use 'shorttitlepage' @tex {\begingroup% \hbox{}\vskip 1.5in \chaprm \centerline{An Introduction to}% \endgroup}% {\begingroup\hbox{}\vskip 0.25in \chaprm% \centerline{Programming in Emacs Lisp}% \endgroup\page\hbox{}\page} @end tex @titlepage @sp 6 @center @titlefont{An Introduction to} @sp 2 @center @titlefont{Programming in Emacs Lisp} @sp 2 @ifset titlepage-edition-number @center @value{titlepage-edition-number} @end ifset @sp 4 @center by Robert J. Chassell @page @vskip 0pt plus 1filll @insertcopying @end titlepage @iftex @headings off @evenheading @thispage @| @| @thischapter @oddheading @thissection @| @| @thispage @end iftex @ifnothtml @c Keep T.O.C. short by tightening up for largebook @ifset largebook @tex \global\parskip 2pt plus 1pt \global\advance\baselineskip by -1pt @end tex @end ifset @end ifnothtml @c If you think this manual is too large for an introduction, please @c consider this email exchange: @c @c >> The intro is almost 300 pages in full. I had expected 60 pages. @c > @c > This is an important point in its own right. Could you @c > write a simplified introduction that is only 50 pages or so? @c > That would be helpful to many potential users, I'd think. @c @c > The problem with the introduction is that it was written when @c > programming was only starting to be a skill "normal" people could @c > have access to. So the text is extremely verbose and is @c > sometimes hard to follow because of that. The gist of the @c > document could be summarized in 50 pages. @c @c This book is intentionally addressed to people who don't know how to @c program. That is its purpose. We recommend people start learning to @c program using this book. @c @c If you DO know how to program in some other language, you can probably @c learn Emacs Lisp starting with the Emacs Lisp Reference Manual. @c @c Richard Stallman , @c https://lists.gnu.org/archive/html/emacs-devel/2018-05/msg00374.html @shortcontents @contents @ifnottex @node Top @top An Introduction to Programming in Emacs Lisp @ifset WWW_GNU_ORG @html

The homepage for GNU Emacs is at https://www.gnu.org/software/emacs/.
To view this manual in other formats, click here. @end html @end ifset @insertcopying This master menu first lists each chapter and index; then it lists every node in every chapter. @end ifnottex @c Uncomment the 3 lines below, starting with @iftex, if you want the @c pages of Preface to be numbered in roman numerals. Use -9 instead @c of -11 for smallbook format. @c >>>> Set pageno appropriately <<<< @c The first page of the Preface is a roman numeral; it is the first @c right handed page after the Table of Contents; hence the following @c setting must be for an odd negative number. @c iftex @c global@pageno = -11 @c end iftex @set COUNT-WORDS count-words-example @c Length of variable name chosen so that things still line up when expanded. @menu * Preface:: What to look for. * List Processing:: What is Lisp? * Practicing Evaluation:: Running several programs. * Writing Defuns:: How to write function definitions. * Buffer Walk Through:: Exploring a few buffer-related functions. * More Complex:: A few, even more complex functions. * Narrowing & Widening:: Restricting your and Emacs attention to a region. * car cdr & cons:: Fundamental functions in Lisp. * Cutting & Storing Text:: Removing text and saving it. * List Implementation:: How lists are implemented in the computer. * Yanking:: Pasting stored text. * Loops & Recursion:: How to repeat a process. * Regexp Search:: Regular expression searches. * Counting Words:: A review of repetition and regexps. * Words in a defun:: Counting words in a @code{defun}. * Readying a Graph:: A prototype graph printing function. * Emacs Initialization:: How to write a @file{.emacs} file. * Debugging:: How to run the Emacs Lisp debuggers. * Conclusion:: Now you have the basics. * the-the:: An appendix: how to find reduplicated words. * Kill Ring:: An appendix: how the kill ring works. * Full Graph:: How to create a graph with labeled axes. * Free Software and Free Manuals:: * GNU Free Documentation License:: * Index:: * About the Author:: @detailmenu --- The Detailed Node Listing --- Preface * Why:: Why learn Emacs Lisp? * On Reading this Text:: Read, gain familiarity, pick up habits.... * Who You Are:: For whom this is written. * Lisp History:: * Note for Novices:: You can read this as a novice. * Thank You:: List Processing * Lisp Lists:: What are lists? * Run a Program:: Any list in Lisp is a program ready to run. * Making Errors:: Generating an error message. * Names & Definitions:: Names of symbols and function definitions. * Lisp Interpreter:: What the Lisp interpreter does. * Evaluation:: Running a program. * Variables:: Returning a value from a variable. * Arguments:: Passing information to a function. * set & setq:: Setting the value of a variable. * Summary:: The major points. * Error Message Exercises:: Lisp Lists * Numbers Lists:: List have numbers, other lists, in them. * Lisp Atoms:: Elemental entities. * Whitespace in Lists:: Formatting lists to be readable. * Typing Lists:: How GNU Emacs helps you type lists. The Lisp Interpreter * Complications:: Variables, Special forms, Lists within. * Byte Compiling:: Specially processing code for speed. Evaluation * How the Interpreter Acts:: Returns and Side Effects... * Evaluating Inner Lists:: Lists within lists... Variables * fill-column Example:: * Void Function:: The error message for a symbol without a function. * Void Variable:: The error message for a symbol without a value. Arguments * Data types:: Types of data passed to a function. * Args as Variable or List:: An argument can be the value of a variable or list. * Variable Number of Arguments:: Some functions may take a variable number of arguments. * Wrong Type of Argument:: Passing an argument of the wrong type to a function. * message:: A useful function for sending messages. Setting the Value of a Variable * Using set:: Setting values. * Using setq:: Setting a quoted value. * Counting:: Using @code{setq} to count. Practicing Evaluation * How to Evaluate:: Typing editing commands or @kbd{C-x C-e} causes evaluation. * Buffer Names:: Buffers and files are different. * Getting Buffers:: Getting a buffer itself, not merely its name. * Switching Buffers:: How to change to another buffer. * Buffer Size & Locations:: Where point is located and the size of the buffer. * Evaluation Exercise:: How To Write Function Definitions * Primitive Functions:: * defun:: The @code{defun} macro. * Install:: Install a function definition. * Interactive:: Making a function interactive. * Interactive Options:: Different options for @code{interactive}. * Permanent Installation:: Installing code permanently. * let:: Creating and initializing local variables. * if:: What if? * else:: If--then--else expressions. * Truth & Falsehood:: What Lisp considers false and true. * save-excursion:: Keeping track of point and buffer. * Review:: * defun Exercises:: Install a Function Definition * Effect of installation:: * Change a defun:: How to change a function definition. Make a Function Interactive * Interactive multiply-by-seven:: An overview. * multiply-by-seven in detail:: The interactive version. @code{let} * Prevent confusion:: * Parts of let Expression:: * Sample let Expression:: * Uninitialized let Variables:: The @code{if} Special Form * if in more detail:: * type-of-animal in detail:: An example of an @code{if} expression. Truth and Falsehood in Emacs Lisp * nil explained:: @code{nil} has two meanings. @code{save-excursion} * Point and mark:: A review of various locations. * Template for save-excursion:: A Few Buffer-Related Functions * Finding More:: How to find more information. * simplified-beginning-of-buffer:: Shows @code{goto-char}, @code{point-min}, and @code{push-mark}. * mark-whole-buffer:: Almost the same as @code{beginning-of-buffer}. * append-to-buffer:: Uses @code{save-excursion} and @code{insert-buffer-substring}. * Buffer Related Review:: Review. * Buffer Exercises:: The Definition of @code{mark-whole-buffer} * mark-whole-buffer overview:: * Body of mark-whole-buffer:: Only three lines of code. The Definition of @code{append-to-buffer} * append-to-buffer overview:: * append interactive:: A two part interactive expression. * append-to-buffer body:: Incorporates a @code{let} expression. * append save-excursion:: How the @code{save-excursion} works. A Few More Complex Functions * copy-to-buffer:: With @code{set-buffer}, @code{get-buffer-create}. * insert-buffer:: Read-only, and with @code{or}. * beginning-of-buffer:: Shows @code{goto-char}, @code{point-min}, and @code{push-mark}. * Second Buffer Related Review:: * optional Exercise:: The Definition of @code{insert-buffer} * insert-buffer code:: * insert-buffer interactive:: When you can read, but not write. * insert-buffer body:: The body has an @code{or} and a @code{let}. * if & or:: Using an @code{if} instead of an @code{or}. * Insert or:: How the @code{or} expression works. * Insert let:: Two @code{save-excursion} expressions. * New insert-buffer:: The Interactive Expression in @code{insert-buffer} * Read-only buffer:: When a buffer cannot be modified. * b for interactive:: An existing buffer or else its name. Complete Definition of @code{beginning-of-buffer} * Optional Arguments:: * beginning-of-buffer opt arg:: Example with optional argument. * beginning-of-buffer complete:: @code{beginning-of-buffer} with an Argument * Disentangle beginning-of-buffer:: * Large buffer case:: * Small buffer case:: Narrowing and Widening * Narrowing advantages:: The advantages of narrowing * save-restriction:: The @code{save-restriction} special form. * what-line:: The number of the line that point is on. * narrow Exercise:: @code{car}, @code{cdr}, @code{cons}: Fundamental Functions * Strange Names:: A historical aside: why the strange names? * car & cdr:: Functions for extracting part of a list. * cons:: Constructing a list. * nthcdr:: Calling @code{cdr} repeatedly. * nth:: * setcar:: Changing the first element of a list. * setcdr:: Changing the rest of a list. * cons Exercise:: @code{cons} * Build a list:: * length:: How to find the length of a list. Cutting and Storing Text * Storing Text:: Text is stored in a list. * zap-to-char:: Cutting out text up to a character. * kill-region:: Cutting text out of a region. * copy-region-as-kill:: A definition for copying text. * Digression into C:: Minor note on C programming language macros. * defvar:: How to give a variable an initial value. * cons & search-fwd Review:: * search Exercises:: @code{zap-to-char} * Complete zap-to-char:: The complete implementation. * zap-to-char interactive:: A three part interactive expression. * zap-to-char body:: A short overview. * search-forward:: How to search for a string. * progn:: The @code{progn} special form. * Summing up zap-to-char:: Using @code{point} and @code{search-forward}. @code{kill-region} * Complete kill-region:: The function definition. * condition-case:: Dealing with a problem. * Lisp macro:: @code{copy-region-as-kill} * Complete copy-region-as-kill:: The complete function definition. * copy-region-as-kill body:: The body of @code{copy-region-as-kill}. The Body of @code{copy-region-as-kill} * last-command & this-command:: * kill-append function:: * kill-new function:: Initializing a Variable with @code{defvar} * See variable current value:: * defvar and asterisk:: How Lists are Implemented * Lists diagrammed:: * Symbols as Chest:: Exploring a powerful metaphor. * List Exercise:: Yanking Text Back * Kill Ring Overview:: * kill-ring-yank-pointer:: The kill ring is a list. * yank nthcdr Exercises:: The @code{kill-ring-yank-pointer} variable. Loops and Recursion * while:: Causing a stretch of code to repeat. * dolist dotimes:: * Recursion:: Causing a function to call itself. * Looping exercise:: @code{while} * Looping with while:: Repeat so long as test returns true. * Loop Example:: A @code{while} loop that uses a list. * print-elements-of-list:: Uses @code{while}, @code{car}, @code{cdr}. * Incrementing Loop:: A loop with an incrementing counter. * Incrementing Loop Details:: * Decrementing Loop:: A loop with a decrementing counter. Details of an Incrementing Loop * Incrementing Example:: Counting pebbles in a triangle. * Inc Example parts:: The parts of the function definition. * Inc Example altogether:: Putting the function definition together. Loop with a Decrementing Counter * Decrementing Example:: More pebbles on the beach. * Dec Example parts:: The parts of the function definition. * Dec Example altogether:: Putting the function definition together. Save your time: @code{dolist} and @code{dotimes} * dolist:: * dotimes:: Recursion * Building Robots:: Same model, different serial number ... * Recursive Definition Parts:: Walk until you stop ... * Recursion with list:: Using a list as the test whether to recurse. * Recursive triangle function:: * Recursion with cond:: * Recursive Patterns:: Often used templates. * No Deferment:: Don't store up work ... * No deferment solution:: Recursion in Place of a Counter * Recursive Example arg of 1 or 2:: * Recursive Example arg of 3 or 4:: Recursive Patterns * Every:: * Accumulate:: * Keep:: Regular Expression Searches * sentence-end:: The regular expression for @code{sentence-end}. * re-search-forward:: Very similar to @code{search-forward}. * forward-sentence:: A straightforward example of regexp search. * forward-paragraph:: A somewhat complex example. * Regexp Review:: * re-search Exercises:: @code{forward-sentence} * Complete forward-sentence:: * fwd-sentence while loops:: Two @code{while} loops. * fwd-sentence re-search:: A regular expression search. @code{forward-paragraph}: a Goldmine of Functions * forward-paragraph in brief:: Key parts of the function definition. * fwd-para let:: The @code{let*} expression. * fwd-para while:: The forward motion @code{while} loop. Counting: Repetition and Regexps * Why Count Words:: * @value{COUNT-WORDS}:: Use a regexp, but find a problem. * recursive-count-words:: Start with case of no words in region. * Counting Exercise:: The @code{@value{COUNT-WORDS}} Function * Design @value{COUNT-WORDS}:: The definition using a @code{while} loop. * Whitespace Bug:: The Whitespace Bug in @code{@value{COUNT-WORDS}}. Counting Words in a @code{defun} * Divide and Conquer:: * Words and Symbols:: What to count? * Syntax:: What constitutes a word or symbol? * count-words-in-defun:: Very like @code{@value{COUNT-WORDS}}. * Several defuns:: Counting several defuns in a file. * Find a File:: Do you want to look at a file? * lengths-list-file:: A list of the lengths of many definitions. * Several files:: Counting in definitions in different files. * Several files recursively:: Recursively counting in different files. * Prepare the data:: Prepare the data for display in a graph. Count Words in @code{defuns} in Different Files * lengths-list-many-files:: Return a list of the lengths of defuns. * append:: Attach one list to another. Prepare the Data for Display in a Graph * Data for Display in Detail:: * Sorting:: Sorting lists. * Files List:: Making a list of files. * Counting function definitions:: Readying a Graph * Columns of a graph:: * graph-body-print:: How to print the body of a graph. * recursive-graph-body-print:: * Printed Axes:: * Line Graph Exercise:: Your @file{.emacs} File * Default Configuration:: * Site-wide Init:: You can write site-wide init files. * defcustom:: Emacs will write code for you. * Beginning init File:: How to write a @file{.emacs} init file. * Text and Auto-fill:: Automatically wrap lines. * Mail Aliases:: Use abbreviations for email addresses. * Indent Tabs Mode:: Don't use tabs with @TeX{} * Keybindings:: Create some personal keybindings. * Keymaps:: More about key binding. * Loading Files:: Load (i.e., evaluate) files automatically. * Autoload:: Make functions available. * Simple Extension:: Define a function; bind it to a key. * X11 Colors:: Colors in X. * Miscellaneous:: * Mode Line:: How to customize your mode line. Debugging * debug:: How to use the built-in debugger. * debug-on-entry:: Start debugging when you call a function. * debug-on-quit:: Start debugging when you quit with @kbd{C-g}. * edebug:: How to use Edebug, a source level debugger. * Debugging Exercises:: Handling the Kill Ring * What the Kill Ring Does:: * current-kill:: * yank:: Paste a copy of a clipped element. * yank-pop:: Insert element pointed to. * ring file:: The @code{current-kill} Function * Code for current-kill:: * Understanding current-kill:: @code{current-kill} in Outline * Body of current-kill:: * Digression concerning error:: How to mislead humans, but not computers. * Determining the Element:: A Graph with Labeled Axes * Labeled Example:: * print-graph Varlist:: @code{let} expression in @code{print-graph}. * print-Y-axis:: Print a label for the vertical axis. * print-X-axis:: Print a horizontal label. * Print Whole Graph:: The function to print a complete graph. The @code{print-Y-axis} Function * print-Y-axis in Detail:: * Height of label:: What height for the Y axis? * Compute a Remainder:: How to compute the remainder of a division. * Y Axis Element:: Construct a line for the Y axis. * Y-axis-column:: Generate a list of Y axis labels. * print-Y-axis Penultimate:: A not quite final version. The @code{print-X-axis} Function * Similarities differences:: Much like @code{print-Y-axis}, but not exactly. * X Axis Tic Marks:: Create tic marks for the horizontal axis. Printing the Whole Graph * The final version:: A few changes. * Test print-graph:: Run a short test. * Graphing words in defuns:: Executing the final code. * lambda:: How to write an anonymous function. * mapcar:: Apply a function to elements of a list. * Another Bug:: Yet another bug @dots{} most insidious. * Final printed graph:: The graph itself! @end detailmenu @end menu @node Preface @unnumbered Preface Most of the GNU Emacs integrated environment is written in the programming language called Emacs Lisp. The code written in this programming language is the software---the sets of instructions---that tell the computer what to do when you give it commands. Emacs is designed so that you can write new code in Emacs Lisp and easily install it as an extension to the editor. (GNU Emacs is sometimes called an ``extensible editor'', but it does much more than provide editing capabilities. It is better to refer to Emacs as an ``extensible computing environment''. However, that phrase is quite a mouthful. It is easier to refer to Emacs simply as an editor. Moreover, everything you do in Emacs---find the Mayan date and phases of the moon, simplify polynomials, debug code, manage files, read letters, write books---all these activities are kinds of editing in the most general sense of the word.) @menu * Why:: Why learn Emacs Lisp? * On Reading this Text:: Read, gain familiarity, pick up habits.... * Who You Are:: For whom this is written. * Lisp History:: * Note for Novices:: You can read this as a novice. * Thank You:: @end menu @ifnottex @node Why @unnumberedsec Why Study Emacs Lisp? @end ifnottex Although Emacs Lisp is usually thought of in association only with Emacs, it is a full computer programming language. You can use Emacs Lisp as you would any other programming language. Perhaps you want to understand programming; perhaps you want to extend Emacs; or perhaps you want to become a programmer. This introduction to Emacs Lisp is designed to get you started: to guide you in learning the fundamentals of programming, and more importantly, to show you how you can teach yourself to go further. @node On Reading this Text @unnumberedsec On Reading this Text All through this document, you will see little sample programs you can run inside of Emacs. If you read this document in Info inside of GNU Emacs, you can run the programs as they appear. (This is easy to do and is explained when the examples are presented.) Alternatively, you can read this introduction as a printed book while sitting beside a computer running Emacs. (This is what I like to do; I like printed books.) If you don't have a running Emacs beside you, you can still read this book, but in this case, it is best to treat it as a novel or as a travel guide to a country not yet visited: interesting, but not the same as being there. Much of this introduction is dedicated to walkthroughs or guided tours of code used in GNU Emacs. These tours are designed for two purposes: first, to give you familiarity with real, working code (code you use every day); and, second, to give you familiarity with the way Emacs works. It is interesting to see how a working environment is implemented. Also, I hope that you will pick up the habit of browsing through source code. You can learn from it and mine it for ideas. Having GNU Emacs is like having a dragon's cave of treasures. In addition to learning about Emacs as an editor and Emacs Lisp as a programming language, the examples and guided tours will give you an opportunity to get acquainted with Emacs as a Lisp programming environment. GNU Emacs supports programming and provides tools that you will want to become comfortable using, such as @kbd{M-.} (the key which invokes the @code{xref-find-definitions} command). You will also learn about buffers and other objects that are part of the environment. Learning about these features of Emacs is like learning new routes around your home town. @ignore In addition, I have written several programs as extended examples. Although these are examples, the programs are real. I use them. Other people use them. You may use them. Beyond the fragments of programs used for illustrations, there is very little in here that is just for teaching purposes; what you see is used. This is a great advantage of Emacs Lisp: it is easy to learn to use it for work. @end ignore Finally, I hope to convey some of the skills for using Emacs to learn aspects of programming that you don't know. You can often use Emacs to help you understand what puzzles you or to find out how to do something new. This self-reliance is not only a pleasure, but an advantage. @node Who You Are @unnumberedsec For Whom This is Written This text is written as an elementary introduction for people who are not programmers. If you are a programmer, you may not be satisfied with this primer. The reason is that you may have become expert at reading reference manuals and be put off by the way this text is organized. An expert programmer who reviewed this text said to me: @quotation @i{I prefer to learn from reference manuals. I ``dive into'' each paragraph, and ``come up for air'' between paragraphs.} @i{When I get to the end of a paragraph, I assume that subject is done, finished, that I know everything I need (with the possible exception of the case when the next paragraph starts talking about it in more detail). I expect that a well written reference manual will not have a lot of redundancy, and that it will have excellent pointers to the (one) place where the information I want is.} @end quotation This introduction is not written for this person! Firstly, I try to say everything at least three times: first, to introduce it; second, to show it in context; and third, to show it in a different context, or to review it. Secondly, I hardly ever put all the information about a subject in one place, much less in one paragraph. To my way of thinking, that imposes too heavy a burden on the reader. Instead I try to explain only what you need to know at the time. (Sometimes I include a little extra information so you won't be surprised later when the additional information is formally introduced.) When you read this text, you are not expected to learn everything the first time. Frequently, you need make only a nodding acquaintance with some of the items mentioned. My hope is that I have structured the text and given you enough hints that you will be alert to what is important, and concentrate on it. You will need to dive into some paragraphs; there is no other way to read them. But I have tried to keep down the number of such paragraphs. This book is intended as an approachable hill, rather than as a daunting mountain. This book, @cite{An Introduction to Programming in Emacs Lisp}, has a companion document, @iftex @cite{The GNU Emacs Lisp Reference Manual}. @end iftex @ifnottex @ref{Top, , The GNU Emacs Lisp Reference Manual, elisp, The GNU Emacs Lisp Reference Manual}. @end ifnottex The reference manual has more detail than this introduction. In the reference manual, all the information about one topic is concentrated in one place. You should turn to it if you are like the programmer quoted above. And, of course, after you have read this @cite{Introduction}, you will find the @cite{Reference Manual} useful when you are writing your own programs. @node Lisp History @unnumberedsec Lisp History @cindex Lisp history Lisp was first developed in the late 1950s at the Massachusetts Institute of Technology for research in artificial intelligence. The great power of the Lisp language makes it superior for other purposes as well, such as writing editor commands and integrated environments. @cindex Maclisp @cindex Common Lisp GNU Emacs Lisp is largely inspired by Maclisp, which was written at MIT in the 1960s. It is somewhat inspired by Common Lisp, which became a standard in the 1980s. However, Emacs Lisp is much simpler than Common Lisp. (The standard Emacs distribution contains an optional extensions file, @file{cl.el}, that adds many Common Lisp features to Emacs Lisp.) @node Note for Novices @unnumberedsec A Note for Novices If you don't know GNU Emacs, you can still read this document profitably. However, I recommend you learn Emacs, if only to learn to move around your computer screen. You can teach yourself how to use Emacs with the built-in tutorial. To use it, type @kbd{C-h t}. (This means you press and release the @key{CTRL} key and the @kbd{h} at the same time, and then press and release @kbd{t}.) Also, I often refer to one of Emacs's standard commands by listing the keys which you press to invoke the command and then giving the name of the command in parentheses, like this: @kbd{M-C-\} (@code{indent-region}). What this means is that the @code{indent-region} command is customarily invoked by typing @kbd{M-C-\}. (You can, if you wish, change the keys that are typed to invoke the command; this is called @dfn{rebinding}. @xref{Keymaps, , Keymaps}.) The abbreviation @kbd{M-C-\} means that you type your @key{META} key, @key{CTRL} key and @kbd{\} key all at the same time. (On many modern keyboards the @key{META} key is labeled @key{ALT}.) Sometimes a combination like this is called a keychord, since it is similar to the way you play a chord on a piano. If your keyboard does not have a @key{META} key, the @key{ESC} key prefix is used in place of it. In this case, @kbd{M-C-\} means that you press and release your @key{ESC} key and then type the @key{CTRL} key and the @kbd{\} key at the same time. But usually @kbd{M-C-\} means press the @key{CTRL} key along with the key that is labeled @key{ALT} and, at the same time, press the @kbd{\} key. In addition to typing a lone keychord, you can prefix what you type with @kbd{C-u}, which is called the @dfn{universal argument}. The @kbd{C-u} keychord passes an argument to the subsequent command. Thus, to indent a region of plain text by 6 spaces, mark the region, and then type @w{@kbd{C-u 6 M-C-\}}. (If you do not specify a number, Emacs either passes the number 4 to the command or otherwise runs the command differently than it would otherwise.) @xref{Arguments, , Numeric Arguments, emacs, The GNU Emacs Manual}. If you are reading this in Info using GNU Emacs, you can read through this whole document just by pressing the space bar, @key{SPC}. (To learn about Info, type @kbd{C-h i} and then select Info.) A note on terminology: when I use the word Lisp alone, I often am referring to the various dialects of Lisp in general, but when I speak of Emacs Lisp, I am referring to GNU Emacs Lisp in particular. @node Thank You @unnumberedsec Thank You My thanks to all who helped me with this book. My especial thanks to @r{Jim Blandy}, @r{Noah Friedman}, @w{Jim Kingdon}, @r{Roland McGrath}, @w{Frank Ritter}, @w{Randy Smith}, @w{Richard M. Stallman}, and @w{Melissa Weisshaus}. My thanks also go to both @w{Philip Johnson} and @w{David Stampe} for their patient encouragement. My mistakes are my own. @flushright Robert J. Chassell @ifnothtml @email{bob@@gnu.org} @end ifnothtml @ifhtml bob@@gnu.org @end ifhtml @end flushright @c ================ Beginning of main text ================ @c Start main text on right-hand (verso) page @tex \par\vfill\supereject \headings off \ifodd\pageno \par\vfill\supereject \else \par\vfill\supereject \page\hbox{}\page \par\vfill\supereject \fi @end tex @c Note: this resetting of the page number back to 1 causes TeX to gripe @c about already having seen page numbers 1-4 before (in the preface): @c pdfTeX warning (ext4): destination with the same identifier (name{1}) @c has been already used, duplicate ignored @c I guess that is harmless (what happens if a later part of the text @c makes a link to something in the first 4 pages though?). @c E.g., note that the Emacs manual has a preface, but does not bother @c resetting the page numbers back to 1 after that. @c Alternatively, uncomment the 3 lines above (search for ``pageno'') @c to have the preface numbered in roman numerals. @iftex @headings off @evenheading @thispage @| @| @thischapter @oddheading @thissection @| @| @thispage @global@pageno = 1 @end iftex @node List Processing @chapter List Processing To the untutored eye, Lisp is a strange programming language. In Lisp code there are parentheses everywhere. Some people even claim that the name stands for ``Lots of Isolated Silly Parentheses''. But the claim is unwarranted. Lisp stands for LISt Processing, and the programming language handles @emph{lists} (and lists of lists) by putting them between parentheses. The parentheses mark the boundaries of the list. Sometimes a list is preceded by an apostrophe @samp{'}, called a @dfn{single-quote} in Lisp.@footnote{A single-quote is an abbreviation for the special form @code{quote}; you need not think about special forms now. @ifnottex @xref{Complications}. @end ifnottex @iftex @xref{Lisp Interpreter}. @end iftex } Lists are the basis of Lisp. @menu * Lisp Lists:: What are lists? * Run a Program:: Any list in Lisp is a program ready to run. * Making Errors:: Generating an error message. * Names & Definitions:: Names of symbols and function definitions. * Lisp Interpreter:: What the Lisp interpreter does. * Evaluation:: Running a program. * Variables:: Returning a value from a variable. * Arguments:: Passing information to a function. * set & setq:: Setting the value of a variable. * Summary:: The major points. * Error Message Exercises:: @end menu @node Lisp Lists @section Lisp Lists @cindex Lisp Lists In Lisp, a list looks like this: @code{'(rose violet daisy buttercup)}. This list is preceded by a single apostrophe. It could just as well be written as follows, which looks more like the kind of list you are likely to be familiar with: @smallexample @group '(rose violet daisy buttercup) @end group @end smallexample @noindent The elements of this list are the names of the four different flowers, separated from each other by whitespace and surrounded by parentheses, like flowers in a field with a stone wall around them. @cindex Flowers in a field @menu * Numbers Lists:: List have numbers, other lists, in them. * Lisp Atoms:: Elemental entities. * Whitespace in Lists:: Formatting lists to be readable. * Typing Lists:: How GNU Emacs helps you type lists. @end menu @ifnottex @node Numbers Lists @unnumberedsubsec Numbers, Lists inside of Lists @end ifnottex Lists can also have numbers in them, as in this list: @code{(+ 2 2)}. This list has a plus-sign, @samp{+}, followed by two @samp{2}s, each separated by whitespace. In Lisp, both data and programs are represented the same way; that is, they are both lists of words, numbers, or other lists, separated by whitespace and surrounded by parentheses. (Since a program looks like data, one program may easily serve as data for another; this is a very powerful feature of Lisp.) (Incidentally, these two parenthetical remarks are @emph{not} Lisp lists, because they contain @samp{;} and @samp{.} as punctuation marks.) @need 1200 Here is another list, this time with a list inside of it: @smallexample '(this list has (a list inside of it)) @end smallexample The components of this list are the words @samp{this}, @samp{list}, @samp{has}, and the list @samp{(a list inside of it)}. The interior list is made up of the words @samp{a}, @samp{list}, @samp{inside}, @samp{of}, @samp{it}. @node Lisp Atoms @subsection Lisp Atoms @cindex Lisp Atoms In Lisp, what we have been calling words are called @dfn{atoms}. This term comes from the historical meaning of the word atom, which means ``indivisible''. As far as Lisp is concerned, the words we have been using in the lists cannot be divided into any smaller parts and still mean the same thing as part of a program; likewise with numbers and single character symbols like @samp{+}. On the other hand, unlike an ancient atom, a list can be split into parts. (@xref{car cdr & cons, , @code{car} @code{cdr} & @code{cons} Fundamental Functions}.) In a list, atoms are separated from each other by whitespace. They can be right next to a parenthesis. @cindex @samp{empty list} defined Technically speaking, a list in Lisp consists of parentheses surrounding atoms separated by whitespace or surrounding other lists or surrounding both atoms and other lists. A list can have just one atom in it or have nothing in it at all. A list with nothing in it looks like this: @code{()}, and is called the @dfn{empty list}. Unlike anything else, an empty list is considered both an atom and a list at the same time. @cindex Symbolic expressions, introduced @cindex @samp{expression} defined @cindex @samp{form} defined The printed representation of both atoms and lists are called @dfn{symbolic expressions} or, more concisely, @dfn{s-expressions}. The word @dfn{expression} by itself can refer to either the printed representation, or to the atom or list as it is held internally in the computer. Often, people use the term @dfn{expression} indiscriminately. (Also, in many texts, the word @dfn{form} is used as a synonym for expression.) Incidentally, the atoms that make up our universe were named such when they were thought to be indivisible; but it has been found that physical atoms are not indivisible. Parts can split off an atom or it can fission into two parts of roughly equal size. Physical atoms were named prematurely, before their truer nature was found. In Lisp, certain kinds of atom, such as an array, can be separated into parts; but the mechanism for doing this is different from the mechanism for splitting a list. As far as list operations are concerned, the atoms of a list are unsplittable. As in English, the meanings of the component letters of a Lisp atom are different from the meaning the letters make as a word. For example, the word for the South American sloth, the @samp{ai}, is completely different from the two words, @samp{a}, and @samp{i}. There are many kinds of atom in nature but only a few in Lisp: for example, @dfn{numbers}, such as 37, 511, or 1729, and @dfn{symbols}, such as @samp{+}, @samp{foo}, or @samp{forward-line}. The words we have listed in the examples above are all symbols. In everyday Lisp conversation, the word ``atom'' is not often used, because programmers usually try to be more specific about what kind of atom they are dealing with. Lisp programming is mostly about symbols (and sometimes numbers) within lists. (Incidentally, the preceding three word parenthetical remark is a proper list in Lisp, since it consists of atoms, which in this case are symbols, separated by whitespace and enclosed by parentheses, without any non-Lisp punctuation.) @need 1250 Text between double quotation marks---even sentences or paragraphs---is also an atom. Here is an example: @cindex Text between double quotation marks @smallexample '(this list includes "text between quotation marks.") @end smallexample @cindex @samp{string} defined @noindent In Lisp, all of the quoted text including the punctuation mark and the blank spaces is a single atom. This kind of atom is called a @dfn{string} (for ``string of characters'') and is the sort of thing that is used for messages that a computer can print for a human to read. Strings are a different kind of atom than numbers or symbols and are used differently. @node Whitespace in Lists @subsection Whitespace in Lists @cindex Whitespace in lists @need 1200 The amount of whitespace in a list does not matter. From the point of view of the Lisp language, @smallexample @group '(this list looks like this) @end group @end smallexample @need 800 @noindent is exactly the same as this: @smallexample '(this list looks like this) @end smallexample Both examples show what to Lisp is the same list, the list made up of the symbols @samp{this}, @samp{list}, @samp{looks}, @samp{like}, and @samp{this} in that order. Extra whitespace and newlines are designed to make a list more readable by humans. When Lisp reads the expression, it gets rid of all the extra whitespace (but it needs to have at least one space between atoms in order to tell them apart.) Odd as it seems, the examples we have seen cover almost all of what Lisp lists look like! Every other list in Lisp looks more or less like one of these examples, except that the list may be longer and more complex. In brief, a list is between parentheses, a string is between quotation marks, a symbol looks like a word, and a number looks like a number. (For certain situations, square brackets, dots and a few other special characters may be used; however, we will go quite far without them.) @node Typing Lists @subsection GNU Emacs Helps You Type Lists @cindex Help typing lists @cindex Formatting help When you type a Lisp expression in GNU Emacs using either Lisp Interaction mode or Emacs Lisp mode, you have available to you several commands to format the Lisp expression so it is easy to read. For example, pressing the @key{TAB} key automatically indents the line the cursor is on by the right amount. A command to properly indent the code in a region is customarily bound to @kbd{M-C-\}. Indentation is designed so that you can see which elements of a list belong to which list---elements of a sub-list are indented more than the elements of the enclosing list. In addition, when you type a closing parenthesis, Emacs momentarily jumps the cursor back to the matching opening parenthesis, so you can see which one it is. This is very useful, since every list you type in Lisp must have its closing parenthesis match its opening parenthesis. (@xref{Major Modes, , Major Modes, emacs, The GNU Emacs Manual}, for more information about Emacs's modes.) @node Run a Program @section Run a Program @cindex Run a program @cindex Program, running one @cindex @samp{evaluate} defined A list in Lisp---any list---is a program ready to run. If you run it (for which the Lisp jargon is @dfn{evaluate}), the computer will do one of three things: do nothing except return to you the list itself; send you an error message; or, treat the first symbol in the list as a command to do something. (Usually, of course, it is the last of these three things that you really want!) @c use code for the single apostrophe, not samp. @findex quote @cindex @code{'} for quoting @cindex quoting using apostrophe @cindex apostrophe for quoting The single apostrophe, @code{'}, that I put in front of some of the example lists in preceding sections is called a @dfn{quote}; when it precedes a list, it tells Lisp to do nothing with the list, other than take it as it is written. But if there is no quote preceding a list, the first item of the list is special: it is a command for the computer to obey. (In Lisp, these commands are called @emph{functions}.) The list @code{(+ 2 2)} shown above did not have a quote in front of it, so Lisp understands that the @code{+} is an instruction to do something with the rest of the list: add the numbers that follow. @need 1250 If you are reading this inside of GNU Emacs in Info, here is how you can evaluate such a list: place your cursor immediately after the right hand parenthesis of the following list and then type @kbd{C-x C-e}: @smallexample (+ 2 2) @end smallexample @c use code for the number four, not samp. @noindent You will see the number @code{4} appear in the echo area@footnote{ Emacs shows integer values in decimal, in octal and in hex, and also as a character, but let's ignore this convenience feature for now. }. (What you have just done is evaluate the list. The echo area is the line at the bottom of the screen that displays or echoes text.) Now try the same thing with a quoted list: place the cursor right after the following list and type @kbd{C-x C-e}: @smallexample '(this is a quoted list) @end smallexample @noindent You will see @code{(this is a quoted list)} appear in the echo area. @cindex Lisp interpreter, explained @cindex Interpreter, Lisp, explained In both cases, what you are doing is giving a command to the program inside of GNU Emacs called the @dfn{Lisp interpreter}---giving the interpreter a command to evaluate the expression. The name of the Lisp interpreter comes from the word for the task done by a human who comes up with the meaning of an expression---who interprets it. You can also evaluate an atom that is not part of a list---one that is not surrounded by parentheses; again, the Lisp interpreter translates from the humanly readable expression to the language of the computer. But before discussing this (@pxref{Variables}), we will discuss what the Lisp interpreter does when you make an error. @node Making Errors @section Generate an Error Message @cindex Generate an error message @cindex Error message generation Partly so you won't worry if you do it accidentally, we will now give a command to the Lisp interpreter that generates an error message. This is a harmless activity; and indeed, we will often try to generate error messages intentionally. Once you understand the jargon, error messages can be informative. Instead of being called ``error'' messages, they should be called ``help'' messages. They are like signposts to a traveler in a strange country; deciphering them can be hard, but once understood, they can point the way. The error message is generated by a built-in GNU Emacs debugger. We will enter the debugger. You get out of the debugger by typing @code{q}. What we will do is evaluate a list that is not quoted and does not have a meaningful command as its first element. Here is a list almost exactly the same as the one we just used, but without the single-quote in front of it. Position the cursor right after it and type @kbd{C-x C-e}: @smallexample (this is an unquoted list) @end smallexample @ignore @noindent What you see depends on which version of Emacs you are running. GNU Emacs version 22 provides more information than version 20 and before. First, the more recent result of generating an error; then the earlier, version 20 result. @need 1250 @noindent In GNU Emacs version 22, a @file{*Backtrace*} window will open up and you will see the following in it: @end ignore A @file{*Backtrace*} window will open up and you should see the following in it: @smallexample @group ---------- Buffer: *Backtrace* ---------- Debugger entered--Lisp error: (void-function this) (this is an unquoted list) eval((this is an unquoted list) nil) elisp--eval-last-sexp(nil) eval-last-sexp(nil) funcall-interactively(eval-last-sexp nil) call-interactively(eval-last-sexp nil nil) command-execute(eval-last-sexp) ---------- Buffer: *Backtrace* ---------- @end group @end smallexample @need 1200 @noindent Your cursor will be in this window (you may have to wait a few seconds before it becomes visible). To quit the debugger and make the debugger window go away, type: @smallexample q @end smallexample @noindent Please type @kbd{q} right now, so you become confident that you can get out of the debugger. Then, type @kbd{C-x C-e} again to re-enter it. @cindex @samp{function} defined Based on what we already know, we can almost read this error message. You read the @file{*Backtrace*} buffer from the bottom up; it tells you what Emacs did. When you typed @kbd{C-x C-e}, you made an interactive call to the command @code{eval-last-sexp}. @code{eval} is an abbreviation for ``evaluate'' and @code{sexp} is an abbreviation for ``symbolic expression''. The command means ``evaluate last symbolic expression'', which is the expression just before your cursor. Each line above tells you what the Lisp interpreter evaluated next. The most recent action is at the top. The buffer is called the @file{*Backtrace*} buffer because it enables you to track Emacs backwards. @need 800 At the top of the @file{*Backtrace*} buffer, you see the line: @smallexample Debugger entered--Lisp error: (void-function this) @end smallexample @noindent The Lisp interpreter tried to evaluate the first atom of the list, the word @samp{this}. It is this action that generated the error message @samp{void-function this}. The message contains the words @samp{void-function} and @samp{this}. @cindex @samp{function} defined The word @samp{function} was mentioned once before. It is a very important word. For our purposes, we can define it by saying that a @dfn{function} is a set of instructions to the computer that tell the computer to do something. Now we can begin to understand the error message: @samp{void-function this}. The function (that is, the word @samp{this}) does not have a definition of any set of instructions for the computer to carry out. The slightly odd word, @samp{void-function}, is designed to cover the way Emacs Lisp is implemented, which is that when a symbol does not have a function definition attached to it, the place that should contain the instructions is void. On the other hand, since we were able to add 2 plus 2 successfully, by evaluating @code{(+ 2 2)}, we can infer that the symbol @code{+} must have a set of instructions for the computer to obey and those instructions must be to add the numbers that follow the @code{+}. It is possible to prevent Emacs entering the debugger in cases like this. We do not explain how to do that here, but we will mention what the result looks like, because you may encounter a similar situation if there is a bug in some Emacs code that you are using. In such cases, you will see only one line of error message; it will appear in the echo area and look like this: @smallexample Symbol's function definition is void:@: this @end smallexample @noindent @ignore (Also, your terminal may beep at you---some do, some don't; and others blink. This is just a device to get your attention.) @end ignore The message goes away as soon as you type a key, even just to move the cursor. We know the meaning of the word @samp{Symbol}. It refers to the first atom of the list, the word @samp{this}. The word @samp{function} refers to the instructions that tell the computer what to do. (Technically, the symbol tells the computer where to find the instructions, but this is a complication we can ignore for the moment.) The error message can be understood: @samp{Symbol's function definition is void:@: this}. The symbol (that is, the word @samp{this}) lacks instructions for the computer to carry out. @node Names & Definitions @section Symbol Names and Function Definitions @cindex Symbol names We can articulate another characteristic of Lisp based on what we have discussed so far---an important characteristic: a symbol, like @code{+}, is not itself the set of instructions for the computer to carry out. Instead, the symbol is used, perhaps temporarily, as a way of locating the definition or set of instructions. What we see is the name through which the instructions can be found. Names of people work the same way. I can be referred to as @samp{Bob}; however, I am not the letters @samp{B}, @samp{o}, @samp{b} but am, or was, the consciousness consistently associated with a particular life-form. The name is not me, but it can be used to refer to me. In Lisp, one set of instructions can be attached to several names. For example, the computer instructions for adding numbers can be linked to the symbol @code{plus} as well as to the symbol @code{+} (and are in some dialects of Lisp). Among humans, I can be referred to as @samp{Robert} as well as @samp{Bob} and by other words as well. On the other hand, a symbol can have only one function definition attached to it at a time. Otherwise, the computer would be confused as to which definition to use. If this were the case among people, only one person in the world could be named @samp{Bob}. However, the function definition to which the name refers can be changed readily. (@xref{Install, , Install a Function Definition}.) Since Emacs Lisp is large, it is customary to name symbols in a way that identifies the part of Emacs to which the function belongs. Thus, all the names for functions that deal with Texinfo start with @samp{texinfo-} and those for functions that deal with reading mail start with @samp{rmail-}. @node Lisp Interpreter @section The Lisp Interpreter @cindex Lisp interpreter, what it does @cindex Interpreter, what it does Based on what we have seen, we can now start to figure out what the Lisp interpreter does when we command it to evaluate a list. First, it looks to see whether there is a quote before the list; if there is, the interpreter just gives us the list. On the other hand, if there is no quote, the interpreter looks at the first element in the list and sees whether it has a function definition. If it does, the interpreter carries out the instructions in the function definition. Otherwise, the interpreter prints an error message. This is how Lisp works. Simple. There are added complications which we will get to in a minute, but these are the fundamentals. Of course, to write Lisp programs, you need to know how to write function definitions and attach them to names, and how to do this without confusing either yourself or the computer. @menu * Complications:: Variables, Special forms, Lists within. * Byte Compiling:: Specially processing code for speed. @end menu @ifnottex @node Complications @unnumberedsubsec Complications @end ifnottex Now, for the first complication. In addition to lists, the Lisp interpreter can evaluate a symbol that is not quoted and does not have parentheses around it. The Lisp interpreter will attempt to determine the symbol's value as a @dfn{variable}. This situation is described in the section on variables. (@xref{Variables}.) @cindex Special form The second complication occurs because some functions are unusual and do not work in the usual manner. Those that don't are called @dfn{special forms}. They are used for special jobs, like defining a function, and there are not many of them. In the next few chapters, you will be introduced to several of the more important special forms. As well as special forms, there are also @dfn{macros}. A macro is a construct defined in Lisp, which differs from a function in that it translates a Lisp expression into another expression that is to be evaluated in place of the original expression. (@xref{Lisp macro}.) For the purposes of this introduction, you do not need to worry too much about whether something is a special form, macro, or ordinary function. For example, @code{if} is a special form (@pxref{if}), but @code{when} is a macro (@pxref{Lisp macro}). In earlier versions of Emacs, @code{defun} was a special form, but now it is a macro (@pxref{defun}). It still behaves in the same way. The final complication is this: if the function that the Lisp interpreter is looking at is not a special form, and if it is part of a list, the Lisp interpreter looks to see whether the list has a list inside of it. If there is an inner list, the Lisp interpreter first figures out what it should do with the inside list, and then it works on the outside list. If there is yet another list embedded inside the inner list, it works on that one first, and so on. It always works on the innermost list first. The interpreter works on the innermost list first, to evaluate the result of that list. The result may be used by the enclosing expression. Otherwise, the interpreter works left to right, from one expression to the next. @node Byte Compiling @subsection Byte Compiling @cindex Byte compiling One other aspect of interpreting: the Lisp interpreter is able to interpret two kinds of entity: humanly readable code, on which we will focus exclusively, and specially processed code, called @dfn{byte compiled} code, which is not humanly readable. Byte compiled code runs faster than humanly readable code. You can transform humanly readable code into byte compiled code by running one of the compile commands such as @code{byte-compile-file}. Byte compiled code is usually stored in a file that ends with a @file{.elc} extension rather than a @file{.el} extension. You will see both kinds of file in the @file{emacs/lisp} directory; the files to read are those with @file{.el} extensions. As a practical matter, for most things you might do to customize or extend Emacs, you do not need to byte compile; and I will not discuss the topic here. @xref{Byte Compilation, , Byte Compilation, elisp, The GNU Emacs Lisp Reference Manual}, for a full description of byte compilation. @node Evaluation @section Evaluation @cindex Evaluation When the Lisp interpreter works on an expression, the term for the activity is called @dfn{evaluation}. We say that the interpreter ``evaluates the expression''. I've used this term several times before. The word comes from its use in everyday language, ``to ascertain the value or amount of; to appraise'', according to @cite{Webster's New Collegiate Dictionary}. @menu * How the Interpreter Acts:: Returns and Side Effects... * Evaluating Inner Lists:: Lists within lists... @end menu @ifnottex @node How the Interpreter Acts @unnumberedsubsec How the Lisp Interpreter Acts @end ifnottex @cindex @samp{returned value} explained After evaluating an expression, the Lisp interpreter will most likely @dfn{return} the value that the computer produces by carrying out the instructions it found in the function definition, or perhaps it will give up on that function and produce an error message. (The interpreter may also find itself tossed, so to speak, to a different function or it may attempt to repeat continually what it is doing for ever and ever in an infinite loop. These actions are less common; and we can ignore them.) Most frequently, the interpreter returns a value. @cindex @samp{side effect} defined At the same time the interpreter returns a value, it may do something else as well, such as move a cursor or copy a file; this other kind of action is called a @dfn{side effect}. Actions that we humans think are important, such as printing results, are often side effects to the Lisp interpreter. It is fairly easy to learn to use side effects. In summary, evaluating a symbolic expression most commonly causes the Lisp interpreter to return a value and perhaps carry out a side effect; or else produce an error. @node Evaluating Inner Lists @subsection Evaluating Inner Lists @cindex Inner list evaluation @cindex Evaluating inner lists If evaluation applies to a list that is inside another list, the outer list may use the value returned by the first evaluation as information when the outer list is evaluated. This explains why inner expressions are evaluated first: the values they return are used by the outer expressions. @need 1250 We can investigate this process by evaluating another addition example. Place your cursor after the following expression and type @kbd{C-x C-e}: @smallexample (+ 2 (+ 3 3)) @end smallexample @noindent The number 8 will appear in the echo area. What happens is that the Lisp interpreter first evaluates the inner expression, @code{(+ 3 3)}, for which the value 6 is returned; then it evaluates the outer expression as if it were written @code{(+ 2 6)}, which returns the value 8. Since there are no more enclosing expressions to evaluate, the interpreter prints that value in the echo area. Now it is easy to understand the name of the command invoked by the keystrokes @kbd{C-x C-e}: the name is @code{eval-last-sexp}. The letters @code{sexp} are an abbreviation for ``symbolic expression'', and @code{eval} is an abbreviation for ``evaluate''. The command evaluates the last symbolic expression. As an experiment, you can try evaluating the expression by putting the cursor at the beginning of the next line immediately following the expression, or inside the expression. @need 800 Here is another copy of the expression: @smallexample (+ 2 (+ 3 3)) @end smallexample @noindent If you place the cursor at the beginning of the blank line that immediately follows the expression and type @kbd{C-x C-e}, you will still get the value 8 printed in the echo area. Now try putting the cursor inside the expression. If you put it right after the next to last parenthesis (so it appears to sit on top of the last parenthesis), you will get a 6 printed in the echo area! This is because the command evaluates the expression @code{(+ 3 3)}. Now put the cursor immediately after a number. Type @kbd{C-x C-e} and you will get the number itself. In Lisp, if you evaluate a number, you get the number itself---this is how numbers differ from symbols. If you evaluate a list starting with a symbol like @code{+}, you will get a value returned that is the result of the computer carrying out the instructions in the function definition attached to that name. If a symbol by itself is evaluated, something different happens, as we will see in the next section. @node Variables @section Variables @cindex Variables In Emacs Lisp, a symbol can have a value attached to it just as it can have a function definition attached to it. The two are different. The function definition is a set of instructions that a computer will obey. A value, on the other hand, is something, such as number or a name, that can vary (which is why such a symbol is called a variable). The value of a symbol can be any expression in Lisp, such as a symbol, number, list, or string. A symbol that has a value is often called a @dfn{variable}. A symbol can have both a function definition and a value attached to it at the same time. Or it can have just one or the other. The two are separate. This is somewhat similar to the way the name Cambridge can refer to the city in Massachusetts and have some information attached to the name as well, such as ``great programming center''. @ignore (Incidentally, in Emacs Lisp, a symbol can have two other things attached to it, too: a property list and a documentation string; these are discussed later.) @end ignore Another way to think about this is to imagine a symbol as being a chest of drawers. The function definition is put in one drawer, the value in another, and so on. What is put in the drawer holding the value can be changed without affecting the contents of the drawer holding the function definition, and vice versa. @menu * fill-column Example:: * Void Function:: The error message for a symbol without a function. * Void Variable:: The error message for a symbol without a value. @end menu @ifnottex @node fill-column Example @unnumberedsubsec @code{fill-column}, an Example Variable @end ifnottex @findex fill-column@r{, an example variable} @cindex Example variable, @code{fill-column} @cindex Variable, example of, @code{fill-column} The variable @code{fill-column} illustrates a symbol with a value attached to it: in every GNU Emacs buffer, this symbol is set to some value, usually 72 or 70, but sometimes to some other value. To find the value of this symbol, evaluate it by itself. If you are reading this in Info inside of GNU Emacs, you can do this by putting the cursor after the symbol and typing @kbd{C-x C-e}: @smallexample fill-column @end smallexample @noindent After I typed @kbd{C-x C-e}, Emacs printed the number 72 in my echo area. This is the value for which @code{fill-column} is set for me as I write this. It may be different for you in your Info buffer. Notice that the value returned as a variable is printed in exactly the same way as the value returned by a function carrying out its instructions. From the point of view of the Lisp interpreter, a value returned is a value returned. What kind of expression it came from ceases to matter once the value is known. A symbol can have any value attached to it or, to use the jargon, we can @dfn{bind} the variable to a value: to a number, such as 72; to a string, @code{"such as this"}; to a list, such as @code{(spruce pine oak)}; we can even bind a variable to a function definition. A symbol can be bound to a value in several ways. @xref{set & setq, , Setting the Value of a Variable}, for information about one way to do this. @node Void Function @subsection Error Message for a Symbol Without a Function @cindex Symbol without function error @cindex Error for symbol without function When we evaluated @code{fill-column} to find its value as a variable, we did not place parentheses around the word. This is because we did not intend to use it as a function name. If @code{fill-column} were the first or only element of a list, the Lisp interpreter would attempt to find the function definition attached to it. But @code{fill-column} has no function definition. Try evaluating this: @smallexample (fill-column) @end smallexample @need 1250 @noindent You will create a @file{*Backtrace*} buffer that says: @smallexample @group ---------- Buffer: *Backtrace* ---------- Debugger entered--Lisp error: (void-function fill-column) (fill-column) eval((fill-column) nil) elisp--eval-last-sexp(nil) eval-last-sexp(nil) funcall-interactively(eval-last-sexp nil) call-interactively(eval-last-sexp nil nil) command-execute(eval-last-sexp) ---------- Buffer: *Backtrace* ---------- @end group @end smallexample @noindent (Remember, to quit the debugger and make the debugger window go away, type @kbd{q} in the @file{*Backtrace*} buffer.) @ignore @need 800 In GNU Emacs 20 and before, you will produce an error message that says: @smallexample Symbol's function definition is void:@: fill-column @end smallexample @noindent (The message will go away as soon as you move the cursor or type another key.) @end ignore @node Void Variable @subsection Error Message for a Symbol Without a Value @cindex Symbol without value error @cindex Error for symbol without value If you attempt to evaluate a symbol that does not have a value bound to it, you will receive an error message. You can see this by experimenting with our 2 plus 2 addition. In the following expression, put your cursor right after the @code{+}, before the first number 2, type @kbd{C-x C-e}: @smallexample (+ 2 2) @end smallexample @need 1500 @noindent In GNU Emacs 22, you will create a @file{*Backtrace*} buffer that says: @smallexample @group ---------- Buffer: *Backtrace* ---------- Debugger entered--Lisp error: (void-variable +) eval(+) elisp--eval-last-sexp(nil) eval-last-sexp(nil) funcall-interactively(eval-last-sexp nil) call-interactively(eval-last-sexp nil nil) command-execute(eval-last-sexp) ---------- Buffer: *Backtrace* ---------- @end group @end smallexample @noindent (Again, you can quit the debugger by typing @kbd{q} in the @file{*Backtrace*} buffer.) This backtrace is different from the very first error message we saw, which said, @samp{Debugger entered--Lisp error: (void-function this)}. In this case, the function does not have a value as a variable; while in the other error message, the function (the word @samp{this}) did not have a definition. In this experiment with the @code{+}, what we did was cause the Lisp interpreter to evaluate the @code{+} and look for the value of the variable instead of the function definition. We did this by placing the cursor right after the symbol rather than after the parenthesis of the enclosing list as we did before. As a consequence, the Lisp interpreter evaluated the preceding s-expression, which in this case was @code{+} by itself. Since @code{+} does not have a value bound to it, just the function definition, the error message reported that the symbol's value as a variable was void. @ignore @need 800 In GNU Emacs version 20 and before, your error message will say: @example Symbol's value as variable is void:@: + @end example @noindent The meaning is the same as in GNU Emacs 22. @end ignore @node Arguments @section Arguments @cindex Arguments @cindex Passing information to functions To see how information is passed to functions, let's look again at our old standby, the addition of two plus two. In Lisp, this is written as follows: @smallexample (+ 2 2) @end smallexample If you evaluate this expression, the number 4 will appear in your echo area. What the Lisp interpreter does is add the numbers that follow the @code{+}. @cindex @samp{argument} defined The numbers added by @code{+} are called the @dfn{arguments} of the function @code{+}. These numbers are the information that is given to or @dfn{passed} to the function. The word ``argument'' comes from the way it is used in mathematics and does not refer to a disputation between two people; instead it refers to the information presented to the function, in this case, to the @code{+}. In Lisp, the arguments to a function are the atoms or lists that follow the function. The values returned by the evaluation of these atoms or lists are passed to the function. Different functions require different numbers of arguments; some functions require none at all.@footnote{It is curious to track the path by which the word ``argument'' came to have two different meanings, one in mathematics and the other in everyday English. According to the @cite{Oxford English Dictionary}, the word derives from the Latin for @samp{to make clear, prove}; thus it came to mean, by one thread of derivation, ``the evidence offered as proof'', which is to say, ``the information offered'', which led to its meaning in Lisp. But in the other thread of derivation, it came to mean ``to assert in a manner against which others may make counter assertions'', which led to the meaning of the word as a disputation. (Note here that the English word has two different definitions attached to it at the same time. By contrast, in Emacs Lisp, a symbol cannot have two different function definitions at the same time.)} @menu * Data types:: Types of data passed to a function. * Args as Variable or List:: An argument can be the value of a variable or list. * Variable Number of Arguments:: Some functions may take a variable number of arguments. * Wrong Type of Argument:: Passing an argument of the wrong type to a function. * message:: A useful function for sending messages. @end menu @node Data types @subsection Arguments' Data Types @cindex Data types @cindex Types of data @cindex Arguments' data types The type of data that should be passed to a function depends on what kind of information it uses. The arguments to a function such as @code{+} must have values that are numbers, since @code{+} adds numbers. Other functions use different kinds of data for their arguments. @need 1250 @findex concat For example, the @code{concat} function links together or unites two or more strings of text to produce a string. The arguments are strings. Concatenating the two character strings @code{abc}, @code{def} produces the single string @code{abcdef}. This can be seen by evaluating the following: @smallexample (concat "abc" "def") @end smallexample @noindent The value produced by evaluating this expression is @code{"abcdef"}. @cindex substring A function such as @code{substring} uses both a string and numbers as arguments. The function returns a part of the string, a @dfn{substring} of the first argument. This function takes three arguments. Its first argument is the string of characters, the second and third arguments are numbers that indicate the beginning (inclusive) and end (exclusive) of the substring. The numbers are a count of the number of characters (including spaces and punctuation) from the beginning of the string. Note that the characters in a string are numbered from zero, not one. @need 800 For example, if you evaluate the following: @smallexample (substring "The quick brown fox jumped." 16 19) @end smallexample @noindent you will see @code{"fox"} appear in the echo area. The arguments are the string and the two numbers. Note that the string passed to @code{substring} is a single atom even though it is made up of several words separated by spaces. Lisp counts everything between the two quotation marks as part of the string, including the spaces. You can think of the @code{substring} function as a kind of atom smasher since it takes an otherwise indivisible atom and extracts a part. However, @code{substring} is only able to extract a substring from an argument that is a string, not from another type of atom such as a number or symbol. @node Args as Variable or List @subsection An Argument as the Value of a Variable or List An argument can be a symbol that returns a value when it is evaluated. For example, when the symbol @code{fill-column} by itself is evaluated, it returns a number. This number can be used in an addition. @need 1250 Position the cursor after the following expression and type @kbd{C-x C-e}: @smallexample (+ 2 fill-column) @end smallexample @noindent The value will be a number two more than what you get by evaluating @code{fill-column} alone. For me, this is 74, because my value of @code{fill-column} is 72. As we have just seen, an argument can be a symbol that returns a value when evaluated. In addition, an argument can be a list that returns a value when it is evaluated. For example, in the following expression, the arguments to the function @code{concat} are the strings @w{@code{"The "}} and @w{@code{" red foxes."}} and the list @code{(number-to-string (+ 2 fill-column))}. @c For GNU Emacs 22, need number-to-string @smallexample (concat "The " (number-to-string (+ 2 fill-column)) " red foxes.") @end smallexample @noindent If you evaluate this expression---and if, as with my Emacs, @code{fill-column} evaluates to 72---@code{"The 74 red foxes."} will appear in the echo area. (Note that you must put spaces after the word @samp{The} and before the word @samp{red} so they will appear in the final string. The function @code{number-to-string} converts the integer that the addition function returns to a string. @code{number-to-string} is also known as @code{int-to-string}.) @node Variable Number of Arguments @subsection Variable Number of Arguments @cindex Variable number of arguments @cindex Arguments, variable number of Some functions, such as @code{concat}, @code{+} or @code{*}, take any number of arguments. (The @code{*} is the symbol for multiplication.) This can be seen by evaluating each of the following expressions in the usual way. What you will see in the echo area is printed in this text after @samp{@result{}}, which you may read as ``evaluates to''. @need 1250 In the first set, the functions have no arguments: @smallexample @group (+) @result{} 0 (*) @result{} 1 @end group @end smallexample @need 1250 In this set, the functions have one argument each: @smallexample @group (+ 3) @result{} 3 (* 3) @result{} 3 @end group @end smallexample @need 1250 In this set, the functions have three arguments each: @smallexample @group (+ 3 4 5) @result{} 12 (* 3 4 5) @result{} 60 @end group @end smallexample @node Wrong Type of Argument @subsection Using the Wrong Type Object as an Argument @cindex Wrong type of argument @cindex Argument, wrong type of When a function is passed an argument of the wrong type, the Lisp interpreter produces an error message. For example, the @code{+} function expects the values of its arguments to be numbers. As an experiment we can pass it the quoted symbol @code{hello} instead of a number. Position the cursor after the following expression and type @kbd{C-x C-e}: @smallexample (+ 2 'hello) @end smallexample @noindent When you do this you will generate an error message. What has happened is that @code{+} has tried to add the 2 to the value returned by @code{'hello}, but the value returned by @code{'hello} is the symbol @code{hello}, not a number. Only numbers can be added. So @code{+} could not carry out its addition. @need 1250 You will create and enter a @file{*Backtrace*} buffer that says: @noindent @smallexample @group ---------- Buffer: *Backtrace* ---------- Debugger entered--Lisp error: (wrong-type-argument number-or-marker-p hello) +(2 hello) eval((+ 2 'hello) nil) elisp--eval-last-sexp(t) eval-last-sexp(nil) funcall-interactively(eval-print-last-sexp nil) call-interactively(eval-print-last-sexp nil nil) command-execute(eval-print-last-sexp) ---------- Buffer: *Backtrace* ---------- @end group @end smallexample @need 1250 As usual, the error message tries to be helpful and makes sense after you learn how to read it.@footnote{@code{(quote hello)} is an expansion of the abbreviation @code{'hello}.} The first part of the error message is straightforward; it says @samp{wrong type argument}. Next comes the mysterious jargon word @w{@samp{number-or-marker-p}}. This word is trying to tell you what kind of argument the @code{+} expected. The symbol @code{number-or-marker-p} says that the Lisp interpreter is trying to determine whether the information presented it (the value of the argument) is a number or a marker (a special object representing a buffer position). What it does is test to see whether the @code{+} is being given numbers to add. It also tests to see whether the argument is something called a marker, which is a specific feature of Emacs Lisp. (In Emacs, locations in a buffer are recorded as markers. When the mark is set with the @kbd{C-@@} or @kbd{C-@key{SPC}} command, its position is kept as a marker. The mark can be considered a number---the number of characters the location is from the beginning of the buffer.) In Emacs Lisp, @code{+} can be used to add the numeric value of marker positions as numbers. The @samp{p} of @code{number-or-marker-p} is the embodiment of a practice started in the early days of Lisp programming. The @samp{p} stands for ``predicate''. In the jargon used by the early Lisp researchers, a predicate refers to a function to determine whether some property is true or false. So the @samp{p} tells us that @code{number-or-marker-p} is the name of a function that determines whether it is true or false that the argument supplied is a number or a marker. Other Lisp symbols that end in @samp{p} include @code{zerop}, a function that tests whether its argument has the value of zero, and @code{listp}, a function that tests whether its argument is a list. Finally, the last part of the error message is the symbol @code{hello}. This is the value of the argument that was passed to @code{+}. If the addition had been passed the correct type of object, the value passed would have been a number, such as 37, rather than a symbol like @code{hello}. But then you would not have got the error message. @ignore @need 1250 In GNU Emacs version 20 and before, the echo area displays an error message that says: @smallexample Wrong type argument:@: number-or-marker-p, hello @end smallexample This says, in different words, the same as the top line of the @file{*Backtrace*} buffer. @end ignore @node message @subsection The @code{message} Function @findex message Like @code{+}, the @code{message} function takes a variable number of arguments. It is used to send messages to the user and is so useful that we will describe it here. @need 1250 A message is printed in the echo area. For example, you can print a message in your echo area by evaluating the following list: @smallexample (message "This message appears in the echo area!") @end smallexample The whole string between double quotation marks is a single argument and is printed @i{in toto}. (Note that in this example, the message itself will appear in the echo area within double quotes; that is because you see the value returned by the @code{message} function. In most uses of @code{message} in programs that you write, the text will be printed in the echo area as a side-effect, without the quotes. @xref{multiply-by-seven in detail, , @code{multiply-by-seven} in detail}, for an example of this.) However, if there is a @samp{%s} in the quoted string of characters, the @code{message} function does not print the @samp{%s} as such, but looks to the argument that follows the string. It evaluates the second argument and prints the value at the location in the string where the @samp{%s} is. @need 1250 You can see this by positioning the cursor after the following expression and typing @kbd{C-x C-e}: @smallexample (message "The name of this buffer is: %s." (buffer-name)) @end smallexample @noindent In Info, @code{"The name of this buffer is: *info*."} will appear in the echo area. The function @code{buffer-name} returns the name of the buffer as a string, which the @code{message} function inserts in place of @code{%s}. To print a value as an integer, use @samp{%d} in the same way as @samp{%s}. For example, to print a message in the echo area that states the value of the @code{fill-column}, evaluate the following: @smallexample (message "The value of fill-column is %d." fill-column) @end smallexample @noindent On my system, when I evaluate this list, @code{"The value of fill-column is 72."} appears in my echo area@footnote{Actually, you can use @code{%s} to print a number. It is non-specific. @code{%d} prints only the part of a number left of a decimal point, and not anything that is not a number.}. If there is more than one @samp{%s} in the quoted string, the value of the first argument following the quoted string is printed at the location of the first @samp{%s} and the value of the second argument is printed at the location of the second @samp{%s}, and so on. @need 1250 For example, if you evaluate the following, @smallexample @group (message "There are %d %s in the office!" (- fill-column 14) "pink elephants") @end group @end smallexample @noindent a rather whimsical message will appear in your echo area. On my system it says, @code{"There are 58 pink elephants in the office!"}. The expression @code{(- fill-column 14)} is evaluated and the resulting number is inserted in place of the @samp{%d}; and the string in double quotes, @code{"pink elephants"}, is treated as a single argument and inserted in place of the @samp{%s}. (That is to say, a string between double quotes evaluates to itself, like a number.) Finally, here is a somewhat complex example that not only illustrates the computation of a number, but also shows how you can use an expression within an expression to generate the text that is substituted for @samp{%s}: @smallexample @group (message "He saw %d %s" (- fill-column 32) (concat "red " (substring "The quick brown foxes jumped." 16 21) " leaping.")) @end group @end smallexample In this example, @code{message} has three arguments: the string, @code{"He saw %d %s"}, the expression, @code{(- fill-column 32)}, and the expression beginning with the function @code{concat}. The value resulting from the evaluation of @code{(- fill-column 32)} is inserted in place of the @samp{%d}; and the value returned by the expression beginning with @code{concat} is inserted in place of the @samp{%s}. When your fill column is 70 and you evaluate the expression, the message @code{"He saw 38 red foxes leaping."} appears in your echo area. @node set & setq @section Setting the Value of a Variable @cindex Variable, setting value @cindex Setting value of variable @cindex @samp{bind} defined There are several ways by which a variable can be given a value. One of the ways is to use either the function @code{set} or the special form @code{setq}. Another way is to use @code{let} (@pxref{let}). (The jargon for this process is to @dfn{bind} a variable to a value.) The following sections not only describe how @code{set} and @code{setq} work but also illustrate how arguments are passed. @menu * Using set:: Setting values. * Using setq:: Setting a quoted value. * Counting:: Using @code{setq} to count. @end menu @node Using set @subsection Using @code{set} @findex set To set the value of the symbol @code{flowers} to the list @code{'(rose violet daisy buttercup)}, evaluate the following expression by positioning the cursor after the expression and typing @kbd{C-x C-e}. @smallexample (set 'flowers '(rose violet daisy buttercup)) @end smallexample @noindent The list @code{(rose violet daisy buttercup)} will appear in the echo area. This is what is @emph{returned} by the @code{set} function. As a side effect, the symbol @code{flowers} is bound to the list; that is, the symbol @code{flowers}, which can be viewed as a variable, is given the list as its value. (This process, by the way, illustrates how a side effect to the Lisp interpreter, setting the value, can be the primary effect that we humans are interested in. This is because every Lisp function must return a value if it does not get an error, but it will only have a side effect if it is designed to have one.) After evaluating the @code{set} expression, you can evaluate the symbol @code{flowers} and it will return the value you just set. Here is the symbol. Place your cursor after it and type @kbd{C-x C-e}. @smallexample flowers @end smallexample @noindent When you evaluate @code{flowers}, the list @code{(rose violet daisy buttercup)} appears in the echo area. Incidentally, if you evaluate @code{'flowers}, the variable with a quote in front of it, what you will see in the echo area is the symbol itself, @code{flowers}. Here is the quoted symbol, so you can try this: @smallexample 'flowers @end smallexample Note also, that when you use @code{set}, you need to quote both arguments to @code{set}, unless you want them evaluated. Since we do not want either argument evaluated, neither the variable @code{flowers} nor the list @code{(rose violet daisy buttercup)}, both are quoted. (When you use @code{set} without quoting its first argument, the first argument is evaluated before anything else is done. If you did this and @code{flowers} did not have a value already, you would get an error message that the @samp{Symbol's value as variable is void}; on the other hand, if @code{flowers} did return a value after it was evaluated, the @code{set} would attempt to set the value that was returned. There are situations where this is the right thing for the function to do; but such situations are rare.) @node Using setq @subsection Using @code{setq} @findex setq As a practical matter, you almost always quote the first argument to @code{set}. The combination of @code{set} and a quoted first argument is so common that it has its own name: the special form @code{setq}. This special form is just like @code{set} except that the first argument is quoted automatically, so you don't need to type the quote mark yourself. Also, as an added convenience, @code{setq} permits you to set several different variables to different values, all in one expression. To set the value of the variable @code{carnivores} to the list @code{'(lion tiger leopard)} using @code{setq}, the following expression is used: @smallexample (setq carnivores '(lion tiger leopard)) @end smallexample @noindent This is exactly the same as using @code{set} except the first argument is automatically quoted by @code{setq}. (The @samp{q} in @code{setq} means @code{quote}.) @need 1250 With @code{set}, the expression would look like this: @smallexample (set 'carnivores '(lion tiger leopard)) @end smallexample Also, @code{setq} can be used to assign different values to different variables. The first argument is bound to the value of the second argument, the third argument is bound to the value of the fourth argument, and so on. For example, you could use the following to assign a list of trees to the symbol @code{trees} and a list of herbivores to the symbol @code{herbivores}: @smallexample @group (setq trees '(pine fir oak maple) herbivores '(gazelle antelope zebra)) @end group @end smallexample @noindent (The expression could just as well have been on one line, but it might not have fit on a page; and humans find it easier to read nicely formatted lists.) Although I have been using the term ``assign'', there is another way of thinking about the workings of @code{set} and @code{setq}; and that is to say that @code{set} and @code{setq} make the symbol @emph{point} to the list. This latter way of thinking is very common and in forthcoming chapters we shall come upon at least one symbol that has ``pointer'' as part of its name. The name is chosen because the symbol has a value, specifically a list, attached to it; or, expressed another way, the symbol is set to point to the list. @node Counting @subsection Counting @cindex Counting Here is an example that shows how to use @code{setq} in a counter. You might use this to count how many times a part of your program repeats itself. First set a variable to zero; then add one to the number each time the program repeats itself. To do this, you need a variable that serves as a counter, and two expressions: an initial @code{setq} expression that sets the counter variable to zero; and a second @code{setq} expression that increments the counter each time it is evaluated. @smallexample @group (setq counter 0) ; @r{Let's call this the initializer.} (setq counter (+ counter 1)) ; @r{This is the incrementer.} counter ; @r{This is the counter.} @end group @end smallexample @noindent (The text following the @samp{;} are comments. @xref{Change a defun, , Change a Function Definition}.) If you evaluate the first of these expressions, the initializer, @code{(setq counter 0)}, and then evaluate the third expression, @code{counter}, the number @code{0} will appear in the echo area. If you then evaluate the second expression, the incrementer, @code{(setq counter (+ counter 1))}, the counter will get the value 1. So if you again evaluate @code{counter}, the number @code{1} will appear in the echo area. Each time you evaluate the second expression, the value of the counter will be incremented. When you evaluate the incrementer, @code{(setq counter (+ counter 1))}, the Lisp interpreter first evaluates the innermost list; this is the addition. In order to evaluate this list, it must evaluate the variable @code{counter} and the number @code{1}. When it evaluates the variable @code{counter}, it receives its current value. It passes this value and the number @code{1} to the @code{+} which adds them together. The sum is then returned as the value of the inner list and passed to the @code{setq} which sets the variable @code{counter} to this new value. Thus, the value of the variable, @code{counter}, is changed. @node Summary @section Summary Learning Lisp is like climbing a hill in which the first part is the steepest. You have now climbed the most difficult part; what remains becomes easier as you progress onwards. @need 1000 In summary, @itemize @bullet @item Lisp programs are made up of expressions, which are lists or single atoms. @item Lists are made up of zero or more atoms or inner lists, separated by whitespace and surrounded by parentheses. A list can be empty. @item Atoms are multi-character symbols, like @code{forward-paragraph}, single character symbols like @code{+}, strings of characters between double quotation marks, or numbers. @item A number evaluates to itself. @item A string between double quotes also evaluates to itself. @item When you evaluate a symbol by itself, its value is returned. @item When you evaluate a list, the Lisp interpreter looks at the first symbol in the list and then at the function definition bound to that symbol. Then the instructions in the function definition are carried out. @item A single-quote @samp{'} tells the Lisp interpreter that it should return the following expression as written, and not evaluate it as it would if the quote were not there. @item Arguments are the information passed to a function. The arguments to a function are computed by evaluating the rest of the elements of the list of which the function is the first element. @item A function always returns a value when it is evaluated (unless it gets an error); in addition, it may also carry out some action that is a side effect. In many cases, a function's primary purpose is to create a side effect. @end itemize @node Error Message Exercises @section Exercises A few simple exercises: @itemize @bullet @item Generate an error message by evaluating an appropriate symbol that is not within parentheses. @item Generate an error message by evaluating an appropriate symbol that is between parentheses. @item Create a counter that increments by two rather than one. @item Write an expression that prints a message in the echo area when evaluated. @end itemize @node Practicing Evaluation @chapter Practicing Evaluation @cindex Practicing evaluation @cindex Evaluation practice Before learning how to write a function definition in Emacs Lisp, it is useful to spend a little time evaluating various expressions that have already been written. These expressions will be lists with the functions as their first (and often only) element. Since some of the functions associated with buffers are both simple and interesting, we will start with those. In this section, we will evaluate a few of these. In another section, we will study the code of several other buffer-related functions, to see how they were written. @menu * How to Evaluate:: Typing editing commands or @kbd{C-x C-e} causes evaluation. * Buffer Names:: Buffers and files are different. * Getting Buffers:: Getting a buffer itself, not merely its name. * Switching Buffers:: How to change to another buffer. * Buffer Size & Locations:: Where point is located and the size of the buffer. * Evaluation Exercise:: @end menu @ifnottex @node How to Evaluate @unnumberedsec How to Evaluate @end ifnottex @i{Whenever you give an editing command} to Emacs Lisp, such as the command to move the cursor or to scroll the screen, @i{you are evaluating an expression,} the first element of which is a function. @i{This is how Emacs works.} @cindex @samp{interactive function} defined @cindex @samp{command} defined When you type keys, you cause the Lisp interpreter to evaluate an expression and that is how you get your results. Even typing plain text involves evaluating an Emacs Lisp function, in this case, one that uses @code{self-insert-command}, which simply inserts the character you typed. The functions you evaluate by typing keystrokes are called @dfn{interactive} functions, or @dfn{commands}; how you make a function interactive will be illustrated in the chapter on how to write function definitions. @xref{Interactive, , Making a Function Interactive}. In addition to typing keyboard commands, we have seen a second way to evaluate an expression: by positioning the cursor after a list and typing @kbd{C-x C-e}. This is what we will do in the rest of this section. There are other ways to evaluate an expression as well; these will be described as we come to them. Besides being used for practicing evaluation, the functions shown in the next few sections are important in their own right. A study of these functions makes clear the distinction between buffers and files, how to switch to a buffer, and how to determine a location within it. @node Buffer Names @section Buffer Names @findex buffer-name @findex buffer-file-name The two functions, @code{buffer-name} and @code{buffer-file-name}, show the difference between a file and a buffer. When you evaluate the following expression, @code{(buffer-name)}, the name of the buffer appears in the echo area. When you evaluate @code{(buffer-file-name)}, the name of the file to which the buffer refers appears in the echo area. Usually, the name returned by @code{(buffer-name)} is the same as the name of the file to which it refers, and the name returned by @code{(buffer-file-name)} is the full path-name of the file. A file and a buffer are two different entities. A file is information recorded permanently in the computer (unless you delete it). A buffer, on the other hand, is information inside of Emacs that will vanish at the end of the editing session (or when you kill the buffer). Usually, a buffer contains information that you have copied from a file; we say the buffer is @dfn{visiting} that file. This copy is what you work on and modify. Changes to the buffer do not change the file, until you save the buffer. When you save the buffer, the buffer is copied to the file and is thus saved permanently. @need 1250 If you are reading this in Info inside of GNU Emacs, you can evaluate each of the following expressions by positioning the cursor after it and typing @kbd{C-x C-e}. @example @group (buffer-name) (buffer-file-name) @end group @end example @noindent When I do this in Info, the value returned by evaluating @code{(buffer-name)} is @file{"*info*"}, and the value returned by evaluating @code{(buffer-file-name)} is @file{nil}. On the other hand, while I am writing this document, the value returned by evaluating @code{(buffer-name)} is @file{"introduction.texinfo"}, and the value returned by evaluating @code{(buffer-file-name)} is @file{"/gnu/work/intro/introduction.texinfo"}. @cindex @code{nil}, history of word The former is the name of the buffer and the latter is the name of the file. In Info, the buffer name is @file{"*info*"}. Info does not point to any file, so the result of evaluating @code{(buffer-file-name)} is @file{nil}. The symbol @code{nil} is from the Latin word for ``nothing''; in this case, it means that the buffer is not associated with any file. (In Lisp, @code{nil} is also used to mean ``false'' and is a synonym for the empty list, @code{()}.) When I am writing, the name of my buffer is @file{"introduction.texinfo"}. The name of the file to which it points is @file{"/gnu/work/intro/introduction.texinfo"}. (In the expressions, the parentheses tell the Lisp interpreter to treat @w{@code{buffer-name}} and @w{@code{buffer-file-name}} as functions; without the parentheses, the interpreter would attempt to evaluate the symbols as variables. @xref{Variables}.) In spite of the distinction between files and buffers, you will often find that people refer to a file when they mean a buffer and vice versa. Indeed, most people say, ``I am editing a file,'' rather than saying, ``I am editing a buffer which I will soon save to a file.'' It is almost always clear from context what people mean. When dealing with computer programs, however, it is important to keep the distinction in mind, since the computer is not as smart as a person. @cindex Buffer, history of word The word ``buffer'', by the way, comes from the meaning of the word as a cushion that deadens the force of a collision. In early computers, a buffer cushioned the interaction between files and the computer's central processing unit. The drums or tapes that held a file and the central processing unit were pieces of equipment that were very different from each other, working at their own speeds, in spurts. The buffer made it possible for them to work together effectively. Eventually, the buffer grew from being an intermediary, a temporary holding place, to being the place where work is done. This transformation is rather like that of a small seaport that grew into a great city: once it was merely the place where cargo was warehoused temporarily before being loaded onto ships; then it became a business and cultural center in its own right. Not all buffers are associated with files. For example, a @file{*scratch*} buffer does not visit any file. Similarly, a @file{*Help*} buffer is not associated with any file. In the old days, when you lacked a @file{~/.emacs} file and started an Emacs session by typing the command @code{emacs} alone, without naming any files, Emacs started with the @file{*scratch*} buffer visible. Nowadays, you will see a splash screen. You can follow one of the commands suggested on the splash screen, visit a file, or press @kbd{q} to quit the splash screen and reach the @file{*scratch*} buffer. If you switch to the @file{*scratch*} buffer, type @code{(buffer-name)}, position the cursor after it, and then type @kbd{C-x C-e} to evaluate the expression. The name @code{"*scratch*"} will be returned and will appear in the echo area. @code{"*scratch*"} is the name of the buffer. When you type @code{(buffer-file-name)} in the @file{*scratch*} buffer and evaluate that, @code{nil} will appear in the echo area, just as it does when you evaluate @code{(buffer-file-name)} in Info. Incidentally, if you are in the @file{*scratch*} buffer and want the value returned by an expression to appear in the @file{*scratch*} buffer itself rather than in the echo area, type @kbd{C-u C-x C-e} instead of @kbd{C-x C-e}. This causes the value returned to appear after the expression. The buffer will look like this: @smallexample (buffer-name)"*scratch*" @end smallexample @noindent You cannot do this in Info since Info is read-only and it will not allow you to change the contents of the buffer. But you can do this in any buffer you can edit; and when you write code or documentation (such as this book), this feature is very useful. @node Getting Buffers @section Getting Buffers @findex current-buffer @findex other-buffer @cindex Getting a buffer The @code{buffer-name} function returns the @emph{name} of the buffer; to get the buffer @emph{itself}, a different function is needed: the @code{current-buffer} function. If you use this function in code, what you get is the buffer itself. A name and the object or entity to which the name refers are different from each other. You are not your name. You are a person to whom others refer by name. If you ask to speak to George and someone hands you a card with the letters @samp{G}, @samp{e}, @samp{o}, @samp{r}, @samp{g}, and @samp{e} written on it, you might be amused, but you would not be satisfied. You do not want to speak to the name, but to the person to whom the name refers. A buffer is similar: the name of the scratch buffer is @file{*scratch*}, but the name is not the buffer. To get a buffer itself, you need to use a function such as @code{current-buffer}. However, there is a slight complication: if you evaluate @code{current-buffer} in an expression on its own, as we will do here, what you see is a printed representation of the name of the buffer without the contents of the buffer. Emacs works this way for two reasons: the buffer may be thousands of lines long---too long to be conveniently displayed; and, another buffer may have the same contents but a different name, and it is important to distinguish between them. @need 800 Here is an expression containing the function: @smallexample (current-buffer) @end smallexample @noindent If you evaluate this expression in Info in Emacs in the usual way, @file{#} will appear in the echo area. The special format indicates that the buffer itself is being returned, rather than just its name. Incidentally, while you can type a number or symbol into a program, you cannot do that with the printed representation of a buffer: the only way to get a buffer itself is with a function such as @code{current-buffer}. A related function is @code{other-buffer}. This returns the most recently selected buffer other than the one you are in currently, not a printed representation of its name. If you have recently switched back and forth from the @file{*scratch*} buffer, @code{other-buffer} will return that buffer. @need 800 You can see this by evaluating the expression: @smallexample (other-buffer) @end smallexample @noindent You should see @file{#} appear in the echo area, or the name of whatever other buffer you switched back from most recently@footnote{Actually, by default, if the buffer from which you just switched is visible to you in another window, @code{other-buffer} will choose the most recent buffer that you cannot see; this is a subtlety that I often forget.}. @node Switching Buffers @section Switching Buffers @findex switch-to-buffer @findex set-buffer @cindex Switching to a buffer The @code{other-buffer} function actually provides a buffer when it is used as an argument to a function that requires one. We can see this by using @code{other-buffer} and @code{switch-to-buffer} to switch to a different buffer. But first, a brief introduction to the @code{switch-to-buffer} function. When you switched back and forth from Info to the @file{*scratch*} buffer to evaluate @code{(buffer-name)}, you most likely typed @kbd{C-x b} and then typed @file{*scratch*}@footnote{Or rather, to save typing, you probably only typed @kbd{RET} if the default buffer was @file{*scratch*}, or if it was different, then you typed just part of the name, such as @code{*sc}, pressed your @kbd{TAB} key to cause it to expand to the full name, and then typed @kbd{RET}.} when prompted in the minibuffer for the name of the buffer to which you wanted to switch. The keystrokes, @kbd{C-x b}, cause the Lisp interpreter to evaluate the interactive function @code{switch-to-buffer}. As we said before, this is how Emacs works: different keystrokes call or run different functions. For example, @kbd{C-f} calls @code{forward-char}, @kbd{M-e} calls @code{forward-sentence}, and so on. By writing @code{switch-to-buffer} in an expression, and giving it a buffer to switch to, we can switch buffers just the way @kbd{C-x b} does: @smallexample (switch-to-buffer (other-buffer)) @end smallexample @noindent The symbol @code{switch-to-buffer} is the first element of the list, so the Lisp interpreter will treat it as a function and carry out the instructions that are attached to it. But before doing that, the interpreter will note that @code{other-buffer} is inside parentheses and work on that symbol first. @code{other-buffer} is the first (and in this case, the only) element of this list, so the Lisp interpreter calls or runs the function. It returns another buffer. Next, the interpreter runs @code{switch-to-buffer}, passing to it, as an argument, the other buffer, which is what Emacs will switch to. If you are reading this in Info, try this now. Evaluate the expression. (To get back, type @kbd{C-x b @key{RET}}.)@footnote{Remember, this expression will move you to your most recent other buffer that you cannot see. If you really want to go to your most recently selected buffer, even if you can still see it, you need to evaluate the following more complex expression: @smallexample (switch-to-buffer (other-buffer (current-buffer) t)) @end smallexample @c noindent In this case, the first argument to @code{other-buffer} tells it which buffer to skip---the current one---and the second argument tells @code{other-buffer} it is OK to switch to a visible buffer. In regular use, @code{switch-to-buffer} takes you to a buffer not visible in windows since you would most likely use @kbd{C-x o} (@code{other-window}) to go to another visible buffer.} In the programming examples in later sections of this document, you will see the function @code{set-buffer} more often than @code{switch-to-buffer}. This is because of a difference between computer programs and humans: humans have eyes and expect to see the buffer on which they are working on their computer terminals. This is so obvious, it almost goes without saying. However, programs do not have eyes. When a computer program works on a buffer, that buffer does not need to be visible on the screen. @code{switch-to-buffer} is designed for humans and does two different things: it switches the buffer to which Emacs's attention is directed; and it switches the buffer displayed in the window to the new buffer. @code{set-buffer}, on the other hand, does only one thing: it switches the attention of the computer program to a different buffer. The buffer on the screen remains unchanged (of course, normally nothing happens there until the command finishes running). @cindex @samp{call} defined Also, we have just introduced another jargon term, the word @dfn{call}. When you evaluate a list in which the first symbol is a function, you are calling that function. The use of the term comes from the notion of the function as an entity that can do something for you if you call it---just as a plumber is an entity who can fix a leak if you call him or her. @node Buffer Size & Locations @section Buffer Size and the Location of Point @cindex Size of buffer @cindex Buffer size @cindex Point location @cindex Location of point Finally, let's look at several rather simple functions, @code{buffer-size}, @code{point}, @code{point-min}, and @code{point-max}. These give information about the size of a buffer and the location of point within it. The function @code{buffer-size} tells you the size of the current buffer; that is, the function returns a count of the number of characters in the buffer. @smallexample (buffer-size) @end smallexample @noindent You can evaluate this in the usual way, by positioning the cursor after the expression and typing @kbd{C-x C-e}. @cindex @samp{point} defined In Emacs, the current position of the cursor is called @dfn{point}. The expression @code{(point)} returns a number that tells you where the cursor is located as a count of the number of characters from the beginning of the buffer up to point. @need 1250 You can see the character count for point in this buffer by evaluating the following expression in the usual way: @smallexample (point) @end smallexample @noindent As I write this, the value of point is 65724. The @code{point} function is frequently used in some of the examples later in this book. @need 1250 The value of point depends, of course, on its location within the buffer. If you evaluate point in this spot, the number will be larger: @smallexample (point) @end smallexample @noindent For me, the value of point in this location is 66043, which means that there are 319 characters (including spaces) between the two expressions. (Doubtless, you will see different numbers, since I will have edited this since I first evaluated point.) @cindex @samp{narrowing} defined The function @code{point-min} is somewhat similar to @code{point}, but it returns the value of the minimum permissible value of point in the current buffer. This is the number 1 unless @dfn{narrowing} is in effect. (Narrowing is a mechanism whereby you can restrict yourself, or a program, to operations on just a part of a buffer. @xref{Narrowing & Widening, , Narrowing and Widening}.) Likewise, the function @code{point-max} returns the value of the maximum permissible value of point in the current buffer. @node Evaluation Exercise @section Exercise Find a file with which you are working and move towards its middle. Find its buffer name, file name, length, and your position in the file. @node Writing Defuns @chapter How To Write Function Definitions @cindex Definition writing @cindex Function definition writing @cindex Writing a function definition When the Lisp interpreter evaluates a list, it looks to see whether the first symbol on the list has a function definition attached to it; or, put another way, whether the symbol points to a function definition. If it does, the computer carries out the instructions in the definition. A symbol that has a function definition is called, simply, a function (although, properly speaking, the definition is the function and the symbol refers to it.) @menu * Primitive Functions:: * defun:: The @code{defun} macro. * Install:: Install a function definition. * Interactive:: Making a function interactive. * Interactive Options:: Different options for @code{interactive}. * Permanent Installation:: Installing code permanently. * let:: Creating and initializing local variables. * if:: What if? * else:: If--then--else expressions. * Truth & Falsehood:: What Lisp considers false and true. * save-excursion:: Keeping track of point and buffer. * Review:: * defun Exercises:: @end menu @ifnottex @node Primitive Functions @unnumberedsec An Aside about Primitive Functions @end ifnottex @cindex Primitive functions @cindex Functions, primitive @cindex C language primitives @cindex Primitives written in C All functions are defined in terms of other functions, except for a few @dfn{primitive} functions that are written in the C programming language. When you write functions' definitions, you will write them in Emacs Lisp and use other functions as your building blocks. Some of the functions you will use will themselves be written in Emacs Lisp (perhaps by you) and some will be primitives written in C@. The primitive functions are used exactly like those written in Emacs Lisp and behave like them. They are written in C so we can easily run GNU Emacs on any computer that has sufficient power and can run C. Let me re-emphasize this: when you write code in Emacs Lisp, you do not distinguish between the use of functions written in C and the use of functions written in Emacs Lisp. The difference is irrelevant. I mention the distinction only because it is interesting to know. Indeed, unless you investigate, you won't know whether an already-written function is written in Emacs Lisp or C. @node defun @section The @code{defun} Macro @findex defun @cindex @samp{function definition} defined In Lisp, a symbol such as @code{mark-whole-buffer} has code attached to it that tells the computer what to do when the function is called. This code is called the @dfn{function definition} and is created by evaluating a Lisp expression that starts with the symbol @code{defun} (which is an abbreviation for @emph{define function}). In subsequent sections, we will look at function definitions from the Emacs source code, such as @code{mark-whole-buffer}. In this section, we will describe a simple function definition so you can see how it looks. This function definition uses arithmetic because it makes for a simple example. Some people dislike examples using arithmetic; however, if you are such a person, do not despair. Hardly any of the code we will study in the remainder of this introduction involves arithmetic or mathematics. The examples mostly involve text in one way or another. A function definition has up to five parts following the word @code{defun}: @enumerate @item The name of the symbol to which the function definition should be attached. @item A list of the arguments that will be passed to the function. If no arguments will be passed to the function, this is an empty list, @code{()}. @item Documentation describing the function. (Technically optional, but strongly recommended.) @item Optionally, an expression to make the function interactive so you can use it by typing @kbd{M-x} and then the name of the function; or by typing an appropriate key or keychord. @cindex @samp{body} defined @item The code that instructs the computer what to do: the @dfn{body} of the function definition. @end enumerate It is helpful to think of the five parts of a function definition as being organized in a template, with slots for each part: @smallexample @group (defun @var{function-name} (@var{arguments}@dots{}) "@var{optional-documentation}@dots{}" (interactive @var{argument-passing-info}) ; @r{optional} @var{body}@dots{}) @end group @end smallexample As an example, here is the code for a function that multiplies its argument by 7. (This example is not interactive. @xref{Interactive, , Making a Function Interactive}, for that information.) @smallexample @group (defun multiply-by-seven (number) "Multiply NUMBER by seven." (* 7 number)) @end group @end smallexample This definition begins with a parenthesis and the symbol @code{defun}, followed by the name of the function. @cindex @samp{argument list} defined The name of the function is followed by a list that contains the arguments that will be passed to the function. This list is called the @dfn{argument list}. In this example, the list has only one element, the symbol, @code{number}. When the function is used, the symbol will be bound to the value that is used as the argument to the function. Instead of choosing the word @code{number} for the name of the argument, I could have picked any other name. For example, I could have chosen the word @code{multiplicand}. I picked the word ``number'' because it tells what kind of value is intended for this slot; but I could just as well have chosen the word ``multiplicand'' to indicate the role that the value placed in this slot will play in the workings of the function. I could have called it @code{foogle}, but that would have been a bad choice because it would not tell humans what it means. The choice of name is up to the programmer and should be chosen to make the meaning of the function clear. Indeed, you can choose any name you wish for a symbol in an argument list, even the name of a symbol used in some other function: the name you use in an argument list is private to that particular definition. In that definition, the name refers to a different entity than any use of the same name outside the function definition. Suppose you have a nick-name ``Shorty'' in your family; when your family members refer to ``Shorty'', they mean you. But outside your family, in a movie, for example, the name ``Shorty'' refers to someone else. Because a name in an argument list is private to the function definition, you can change the value of such a symbol inside the body of a function without changing its value outside the function. The effect is similar to that produced by a @code{let} expression. (@xref{let, , @code{let}}.) @ignore Note also that we discuss the word ``number'' in two different ways: as a symbol that appears in the code, and as the name of something that will be replaced by a something else during the evaluation of the function. In the first case, @code{number} is a symbol, not a number; it happens that within the function, it is a variable who value is the number in question, but our primary interest in it is as a symbol. On the other hand, when we are talking about the function, our interest is that we will substitute a number for the word @var{number}. To keep this distinction clear, we use different typography for the two circumstances. When we talk about this function, or about how it works, we refer to this number by writing @var{number}. In the function itself, we refer to it by writing @code{number}. @end ignore The argument list is followed by the documentation string that describes the function. This is what you see when you type @w{@kbd{C-h f}} and the name of a function. Incidentally, when you write a documentation string like this, you should make the first line a complete sentence since some commands, such as @code{apropos}, print only the first line of a multi-line documentation string. Also, you should not indent the second line of a documentation string, if you have one, because that looks odd when you use @kbd{C-h f} (@code{describe-function}). The documentation string is optional, but it is so useful, it should be included in almost every function you write. @findex * @r{(multiplication)} The third line of the example consists of the body of the function definition. (Most functions' definitions, of course, are longer than this.) In this function, the body is the list, @code{(* 7 number)}, which says to multiply the value of @var{number} by 7. (In Emacs Lisp, @code{*} is the function for multiplication, just as @code{+} is the function for addition.) When you use the @code{multiply-by-seven} function, the argument @code{number} evaluates to the actual number you want used. Here is an example that shows how @code{multiply-by-seven} is used; but don't try to evaluate this yet! @smallexample (multiply-by-seven 3) @end smallexample @noindent The symbol @code{number}, specified in the function definition in the next section, is bound to the value 3 in the actual use of the function. Note that although @code{number} was inside parentheses in the function definition, the argument passed to the @code{multiply-by-seven} function is not in parentheses. The parentheses are written in the function definition so the computer can figure out where the argument list ends and the rest of the function definition begins. If you evaluate this example, you are likely to get an error message. (Go ahead, try it!) This is because we have written the function definition, but not yet told the computer about the definition---we have not yet loaded the function definition in Emacs. Installing a function is the process that tells the Lisp interpreter the definition of the function. Installation is described in the next section. @node Install @section Install a Function Definition @cindex Install a Function Definition @cindex Definition installation @cindex Function definition installation If you are reading this inside of Info in Emacs, you can try out the @code{multiply-by-seven} function by first evaluating the function definition and then evaluating @code{(multiply-by-seven 3)}. A copy of the function definition follows. Place the cursor after the last parenthesis of the function definition and type @kbd{C-x C-e}. When you do this, @code{multiply-by-seven} will appear in the echo area. (What this means is that when a function definition is evaluated, the value it returns is the name of the defined function.) At the same time, this action installs the function definition. @smallexample @group (defun multiply-by-seven (number) "Multiply NUMBER by seven." (* 7 number)) @end group @end smallexample @noindent By evaluating this @code{defun}, you have just installed @code{multiply-by-seven} in Emacs. The function is now just as much a part of Emacs as @code{forward-word} or any other editing function you use. (@code{multiply-by-seven} will stay installed until you quit Emacs. To reload code automatically whenever you start Emacs, see @ref{Permanent Installation, , Installing Code Permanently}.) @menu * Effect of installation:: * Change a defun:: How to change a function definition. @end menu @ifnottex @node Effect of installation @unnumberedsubsec The effect of installation @end ifnottex You can see the effect of installing @code{multiply-by-seven} by evaluating the following sample. Place the cursor after the following expression and type @kbd{C-x C-e}. The number 21 will appear in the echo area. @smallexample (multiply-by-seven 3) @end smallexample If you wish, you can read the documentation for the function by typing @kbd{C-h f} (@code{describe-function}) and then the name of the function, @code{multiply-by-seven}. When you do this, a @file{*Help*} window will appear on your screen that says: @smallexample @group multiply-by-seven is a Lisp function. (multiply-by-seven NUMBER) Multiply NUMBER by seven. @end group @end smallexample @noindent (To return to a single window on your screen, type @kbd{C-x 1}.) @node Change a defun @subsection Change a Function Definition @cindex Changing a function definition @cindex Function definition, how to change @cindex Definition, how to change If you want to change the code in @code{multiply-by-seven}, just rewrite it. To install the new version in place of the old one, evaluate the function definition again. This is how you modify code in Emacs. It is very simple. As an example, you can change the @code{multiply-by-seven} function to add the number to itself seven times instead of multiplying the number by seven. It produces the same answer, but by a different path. At the same time, we will add a comment to the code; a comment is text that the Lisp interpreter ignores, but that a human reader may find useful or enlightening. The comment is that this is the second version. @smallexample @group (defun multiply-by-seven (number) ; @r{Second version.} "Multiply NUMBER by seven." (+ number number number number number number number)) @end group @end smallexample @cindex Comments in Lisp code The comment follows a semicolon, @samp{;}. In Lisp, everything on a line that follows a semicolon is a comment. The end of the line is the end of the comment. To stretch a comment over two or more lines, begin each line with a semicolon. @xref{Beginning init File, , Beginning a @file{.emacs} File}, and @ref{Comments, , Comments, elisp, The GNU Emacs Lisp Reference Manual}, for more about comments. You can install this version of the @code{multiply-by-seven} function by evaluating it in the same way you evaluated the first function: place the cursor after the last parenthesis and type @kbd{C-x C-e}. In summary, this is how you write code in Emacs Lisp: you write a function; install it; test it; and then make fixes or enhancements and install it again. @node Interactive @section Make a Function Interactive @cindex Interactive functions @findex interactive You make a function interactive by placing a list that begins with the special form @code{interactive} immediately after the documentation. A user can invoke an interactive function by typing @kbd{M-x} and then the name of the function; or by typing the keys to which it is bound, for example, by typing @kbd{C-n} for @code{next-line} or @kbd{C-x h} for @code{mark-whole-buffer}. Interestingly, when you call an interactive function interactively, the value returned is not automatically displayed in the echo area. This is because you often call an interactive function for its side effects, such as moving forward by a word or line, and not for the value returned. If the returned value were displayed in the echo area each time you typed a key, it would be very distracting. @menu * Interactive multiply-by-seven:: An overview. * multiply-by-seven in detail:: The interactive version. @end menu @ifnottex @node Interactive multiply-by-seven @unnumberedsubsec An Interactive @code{multiply-by-seven}, An Overview @end ifnottex Both the use of the special form @code{interactive} and one way to display a value in the echo area can be illustrated by creating an interactive version of @code{multiply-by-seven}. @need 1250 Here is the code: @smallexample @group (defun multiply-by-seven (number) ; @r{Interactive version.} "Multiply NUMBER by seven." (interactive "p") (message "The result is %d" (* 7 number))) @end group @end smallexample @noindent You can install this code by placing your cursor after it and typing @kbd{C-x C-e}. The name of the function will appear in your echo area. Then, you can use this code by typing @kbd{C-u} and a number and then typing @kbd{M-x multiply-by-seven} and pressing @key{RET}. The phrase @samp{The result is @dots{}} followed by the product will appear in the echo area. Speaking more generally, you invoke a function like this in either of two ways: @enumerate @item By typing a prefix argument that contains the number to be passed, and then typing @kbd{M-x} and the name of the function, as with @kbd{C-u 3 M-x forward-sentence}; or, @item By typing whatever key or keychord the function is bound to, as with @kbd{C-u 3 M-e}. @end enumerate @noindent Both the examples just mentioned work identically to move point forward three sentences. (Since @code{multiply-by-seven} is not bound to a key, it could not be used as an example of key binding.) (@xref{Keybindings, , Some Keybindings}, to learn how to bind a command to a key.) A @dfn{prefix argument} is passed to an interactive function by typing the @key{META} key followed by a number, for example, @kbd{M-3 M-e}, or by typing @kbd{C-u} and then a number, for example, @kbd{C-u 3 M-e} (if you type @kbd{C-u} without a number, it defaults to 4). @node multiply-by-seven in detail @subsection An Interactive @code{multiply-by-seven} Let's look at the use of the special form @code{interactive} and then at the function @code{message} in the interactive version of @code{multiply-by-seven}. You will recall that the function definition looks like this: @smallexample @group (defun multiply-by-seven (number) ; @r{Interactive version.} "Multiply NUMBER by seven." (interactive "p") (message "The result is %d" (* 7 number))) @end group @end smallexample In this function, the expression, @code{(interactive "p")}, is a list of two elements. The @code{"p"} tells Emacs to pass the prefix argument to the function and use its value for the argument of the function. @need 1000 The argument will be a number. This means that the symbol @code{number} will be bound to a number in the line: @smallexample (message "The result is %d" (* 7 number)) @end smallexample @need 1250 @noindent For example, if your prefix argument is 5, the Lisp interpreter will evaluate the line as if it were: @smallexample (message "The result is %d" (* 7 5)) @end smallexample @noindent (If you are reading this in GNU Emacs, you can evaluate this expression yourself.) First, the interpreter will evaluate the inner list, which is @code{(* 7 5)}. This returns a value of 35. Next, it will evaluate the outer list, passing the values of the second and subsequent elements of the list to the function @code{message}. As we have seen, @code{message} is an Emacs Lisp function especially designed for sending a one line message to a user. (@xref{message, , The @code{message} function}.) In summary, the @code{message} function prints its first argument in the echo area as is, except for occurrences of @samp{%d} or @samp{%s} (and various other %-sequences which we have not mentioned). When it sees a control sequence, the function looks to the second or subsequent arguments and prints the value of the argument in the location in the string where the control sequence is located. In the interactive @code{multiply-by-seven} function, the control string is @samp{%d}, which requires a number, and the value returned by evaluating @code{(* 7 5)} is the number 35. Consequently, the number 35 is printed in place of the @samp{%d} and the message is @samp{The result is 35}. (Note that when you call the function @code{multiply-by-seven}, the message is printed without quotes, but when you call @code{message}, the text is printed in double quotes. This is because the value returned by @code{message} is what appears in the echo area when you evaluate an expression whose first element is @code{message}; but when embedded in a function, @code{message} prints the text as a side effect without quotes.) @node Interactive Options @section Different Options for @code{interactive} @cindex Options for @code{interactive} @cindex Interactive options In the example, @code{multiply-by-seven} used @code{"p"} as the argument to @code{interactive}. This argument told Emacs to interpret your typing either @kbd{C-u} followed by a number or @key{META} followed by a number as a command to pass that number to the function as its argument. Emacs has more than twenty characters predefined for use with @code{interactive}. In almost every case, one of these options will enable you to pass the right information interactively to a function. (@xref{Interactive Codes, , Code Characters for @code{interactive}, elisp, The GNU Emacs Lisp Reference Manual}.) @need 1250 Consider the function @code{zap-to-char}. Its interactive expression is @c FIXME: the interactive expression of zap-to-char has been changed @c (in 2012-04-10). @smallexample (interactive "p\ncZap to char: ") @end smallexample The first part of the argument to @code{interactive} is @samp{p}, with which you are already familiar. This argument tells Emacs to interpret a prefix, as a number to be passed to the function. You can specify a prefix either by typing @kbd{C-u} followed by a number or by typing @key{META} followed by a number. The prefix is the number of specified characters. Thus, if your prefix is three and the specified character is @samp{x}, then you will delete all the text up to and including the third next @samp{x}. If you do not set a prefix, then you delete all the text up to and including the specified character, but no more. The @samp{c} tells the function the name of the character to which to delete. More formally, a function with two or more arguments can have information passed to each argument by adding parts to the string that follows @code{interactive}. When you do this, the information is passed to each argument in the same order it is specified in the @code{interactive} list. In the string, each part is separated from the next part by a @samp{\n}, which is a newline. For example, you can follow @samp{p} with a @samp{\n} and an @samp{cZap to char:@: }. This causes Emacs to pass the value of the prefix argument (if there is one) and the character. In this case, the function definition looks like the following, where @code{arg} and @code{char} are the symbols to which @code{interactive} binds the prefix argument and the specified character: @smallexample @group (defun @var{name-of-function} (arg char) "@var{documentation}@dots{}" (interactive "p\ncZap to char: ") @var{body-of-function}@dots{}) @end group @end smallexample @noindent (The space after the colon in the prompt makes it look better when you are prompted. @xref{copy-to-buffer, , The Definition of @code{copy-to-buffer}}, for an example.) When a function does not take arguments, @code{interactive} does not require any. Such a function contains the simple expression @code{(interactive)}. The @code{mark-whole-buffer} function is like this. Alternatively, if the special letter-codes are not right for your application, you can pass your own arguments to @code{interactive} as a list. @xref{append-to-buffer, , The Definition of @code{append-to-buffer}}, for an example. @xref{Using Interactive, , Using @code{Interactive}, elisp, The GNU Emacs Lisp Reference Manual}, for a more complete explanation about this technique. @node Permanent Installation @section Install Code Permanently @cindex Install code permanently @cindex Permanent code installation @cindex Code installation When you install a function definition by evaluating it, it will stay installed until you quit Emacs. The next time you start a new session of Emacs, the function will not be installed unless you evaluate the function definition again. At some point, you may want to have code installed automatically whenever you start a new session of Emacs. There are several ways of doing this: @itemize @bullet @item If you have code that is just for yourself, you can put the code for the function definition in your @file{.emacs} initialization file. When you start Emacs, your @file{.emacs} file is automatically evaluated and all the function definitions within it are installed. @xref{Emacs Initialization, , Your @file{.emacs} File}. @item Alternatively, you can put the function definitions that you want installed in one or more files of their own and use the @code{load} function to cause Emacs to evaluate and thereby install each of the functions in the files. @xref{Loading Files, , Loading Files}. @item Thirdly, if you have code that your whole site will use, it is usual to put it in a file called @file{site-init.el} that is loaded when Emacs is built. This makes the code available to everyone who uses your machine. (See the @file{INSTALL} file that is part of the Emacs distribution.) @end itemize Finally, if you have code that everyone who uses Emacs may want, you can post it on a computer network or send a copy to the Free Software Foundation. (When you do this, please license the code and its documentation under a license that permits other people to run, copy, study, modify, and redistribute the code and which protects you from having your work taken from you.) If you send a copy of your code to the Free Software Foundation, and properly protect yourself and others, it may be included in the next release of Emacs. In large part, this is how Emacs has grown over the past years, by donations. @node let @section @code{let} @findex let The @code{let} expression is a special form in Lisp that you will need to use in most function definitions. @code{let} is used to attach or bind a symbol to a value in such a way that the Lisp interpreter will not confuse the variable with a variable of the same name that is not part of the function. To understand why the @code{let} special form is necessary, consider the situation in which you own a home that you generally refer to as ``the house'', as in the sentence, ``The house needs painting.'' If you are visiting a friend and your host refers to ``the house'', he is likely to be referring to @emph{his} house, not yours, that is, to a different house. If your friend is referring to his house and you think he is referring to your house, you may be in for some confusion. The same thing could happen in Lisp if a variable that is used inside of one function has the same name as a variable that is used inside of another function, and the two are not intended to refer to the same value. The @code{let} special form prevents this kind of confusion. @menu * Prevent confusion:: * Parts of let Expression:: * Sample let Expression:: * Uninitialized let Variables:: @end menu @ifnottex @node Prevent confusion @unnumberedsubsec @code{let} Prevents Confusion @end ifnottex @cindex @samp{local variable} defined @cindex @samp{variable, local}, defined The @code{let} special form prevents confusion. @code{let} creates a name for a @dfn{local variable} that overshadows any use of the same name outside the @code{let} expression. This is like understanding that whenever your host refers to ``the house'', he means his house, not yours. (Symbols used in argument lists work the same way. @xref{defun, , The @code{defun} Macro}.) Local variables created by a @code{let} expression retain their value @emph{only} within the @code{let} expression itself (and within expressions called within the @code{let} expression); the local variables have no effect outside the @code{let} expression. Another way to think about @code{let} is that it is like a @code{setq} that is temporary and local. The values set by @code{let} are automatically undone when the @code{let} is finished. The setting only affects expressions that are inside the bounds of the @code{let} expression. In computer science jargon, we would say the binding of a symbol is visible only in functions called in the @code{let} form; in Emacs Lisp, the default scoping is dynamic, not lexical. (The non-default lexical binding is not discussed in this manual.) @code{let} can create more than one variable at once. Also, @code{let} gives each variable it creates an initial value, either a value specified by you, or @code{nil}. (In the jargon, this is binding the variable to the value.) After @code{let} has created and bound the variables, it executes the code in the body of the @code{let}, and returns the value of the last expression in the body, as the value of the whole @code{let} expression. (``Execute'' is a jargon term that means to evaluate a list; it comes from the use of the word meaning ``to give practical effect to'' (@cite{Oxford English Dictionary}). Since you evaluate an expression to perform an action, ``execute'' has evolved as a synonym to ``evaluate''.) @node Parts of let Expression @subsection The Parts of a @code{let} Expression @cindex @code{let} expression, parts of @cindex Parts of @code{let} expression @cindex @samp{varlist} defined A @code{let} expression is a list of three parts. The first part is the symbol @code{let}. The second part is a list, called a @dfn{varlist}, each element of which is either a symbol by itself or a two-element list, the first element of which is a symbol. The third part of the @code{let} expression is the body of the @code{let}. The body usually consists of one or more lists. @need 800 A template for a @code{let} expression looks like this: @smallexample (let @var{varlist} @var{body}@dots{}) @end smallexample @noindent The symbols in the varlist are the variables that are given initial values by the @code{let} special form. Symbols by themselves are given the initial value of @code{nil}; and each symbol that is the first element of a two-element list is bound to the value that is returned when the Lisp interpreter evaluates the second element. Thus, a varlist might look like this: @code{(thread (needles 3))}. In this case, in a @code{let} expression, Emacs binds the symbol @code{thread} to an initial value of @code{nil}, and binds the symbol @code{needles} to an initial value of 3. When you write a @code{let} expression, what you do is put the appropriate expressions in the slots of the @code{let} expression template. If the varlist is composed of two-element lists, as is often the case, the template for the @code{let} expression looks like this: @smallexample @group (let ((@var{variable} @var{value}) (@var{variable} @var{value}) @dots{}) @var{body}@dots{}) @end group @end smallexample @node Sample let Expression @subsection Sample @code{let} Expression @cindex Sample @code{let} expression @cindex @code{let} expression sample The following expression creates and gives initial values to the two variables @code{zebra} and @code{tiger}. The body of the @code{let} expression is a list which calls the @code{message} function. @smallexample @group (let ((zebra "stripes") (tiger "fierce")) (message "One kind of animal has %s and another is %s." zebra tiger)) @end group @end smallexample Here, the varlist is @code{((zebra "stripes") (tiger "fierce"))}. The two variables are @code{zebra} and @code{tiger}. Each variable is the first element of a two-element list and each value is the second element of its two-element list. In the varlist, Emacs binds the variable @code{zebra} to the value @code{"stripes"}@footnote{According to Jared Diamond in @cite{Guns, Germs, and Steel}, ``@dots{} zebras become impossibly dangerous as they grow older'' but the claim here is that they do not become fierce like a tiger. (1997, W. W. Norton and Co., ISBN 0-393-03894-2, page 171)}, and binds the variable @code{tiger} to the value @code{"fierce"}. In this example, both values are strings. The values could just as well have been another list or a symbol. The body of the @code{let} follows after the list holding the variables. In this example, the body is a list that uses the @code{message} function to print a string in the echo area. @need 1500 You may evaluate the example in the usual fashion, by placing the cursor after the last parenthesis and typing @kbd{C-x C-e}. When you do this, the following will appear in the echo area: @smallexample "One kind of animal has stripes and another is fierce." @end smallexample As we have seen before, the @code{message} function prints its first argument, except for @samp{%s}. In this example, the value of the variable @code{zebra} is printed at the location of the first @samp{%s} and the value of the variable @code{tiger} is printed at the location of the second @samp{%s}. @node Uninitialized let Variables @subsection Uninitialized Variables in a @code{let} Statement @cindex Uninitialized @code{let} variables @cindex @code{let} variables uninitialized If you do not bind the variables in a @code{let} statement to specific initial values, they will automatically be bound to an initial value of @code{nil}, as in the following expression: @smallexample @group (let ((birch 3) pine fir (oak 'some)) (message "Here are %d variables with %s, %s, and %s value." birch pine fir oak)) @end group @end smallexample @noindent Here, the varlist is @code{((birch 3) pine fir (oak 'some))}. @need 1250 If you evaluate this expression in the usual way, the following will appear in your echo area: @smallexample "Here are 3 variables with nil, nil, and some value." @end smallexample @noindent In this example, Emacs binds the symbol @code{birch} to the number 3, binds the symbols @code{pine} and @code{fir} to @code{nil}, and binds the symbol @code{oak} to the value @code{some}. Note that in the first part of the @code{let}, the variables @code{pine} and @code{fir} stand alone as atoms that are not surrounded by parentheses; this is because they are being bound to @code{nil}, the empty list. But @code{oak} is bound to @code{some} and so is a part of the list @code{(oak 'some)}. Similarly, @code{birch} is bound to the number 3 and so is in a list with that number. (Since a number evaluates to itself, the number does not need to be quoted. Also, the number is printed in the message using a @samp{%d} rather than a @samp{%s}.) The four variables as a group are put into a list to delimit them from the body of the @code{let}. @node if @section The @code{if} Special Form @findex if @cindex Conditional with @code{if} Another special form is the conditional @code{if}. This form is used to instruct the computer to make decisions. You can write function definitions without using @code{if}, but it is used often enough, and is important enough, to be included here. It is used, for example, in the code for the function @code{beginning-of-buffer}. The basic idea behind an @code{if}, is that @emph{if} a test is true, @emph{then} an expression is evaluated. If the test is not true, the expression is not evaluated. For example, you might make a decision such as, ``if it is warm and sunny, then go to the beach!'' @menu * if in more detail:: * type-of-animal in detail:: An example of an @code{if} expression. @end menu @ifnottex @node if in more detail @unnumberedsubsec @code{if} in more detail @end ifnottex @cindex @samp{if-part} defined @cindex @samp{then-part} defined An @code{if} expression written in Lisp does not use the word ``then''; the test and the action are the second and third elements of the list whose first element is @code{if}. Nonetheless, the test part of an @code{if} expression is often called the @dfn{if-part} and the second argument is often called the @dfn{then-part}. Also, when an @code{if} expression is written, the true-or-false-test is usually written on the same line as the symbol @code{if}, but the action to carry out if the test is true, the then-part, is written on the second and subsequent lines. This makes the @code{if} expression easier to read. @smallexample @group (if @var{true-or-false-test} @var{action-to-carry-out-if-test-is-true}) @end group @end smallexample @noindent The true-or-false-test will be an expression that is evaluated by the Lisp interpreter. Here is an example that you can evaluate in the usual manner. The test is whether the number 5 is greater than the number 4. Since it is, the message @samp{5 is greater than 4!} will be printed. @smallexample @group (if (> 5 4) ; @r{if-part} (message "5 is greater than 4!")) ; @r{then-part} @end group @end smallexample @noindent (The function @code{>} tests whether its first argument is greater than its second argument and returns true if it is.) @findex > @r{(greater than)} Of course, in actual use, the test in an @code{if} expression will not be fixed for all time as it is by the expression @code{(> 5 4)}. Instead, at least one of the variables used in the test will be bound to a value that is not known ahead of time. (If the value were known ahead of time, we would not need to run the test!) For example, the value may be bound to an argument of a function definition. In the following function definition, the character of the animal is a value that is passed to the function. If the value bound to @code{characteristic} is @code{"fierce"}, then the message, @samp{It is a tiger!} will be printed; otherwise, @code{nil} will be returned. @smallexample @group (defun type-of-animal (characteristic) "Print message in echo area depending on CHARACTERISTIC. If the CHARACTERISTIC is the string \"fierce\", then warn of a tiger." (if (equal characteristic "fierce") (message "It is a tiger!"))) @end group @end smallexample @need 1500 @noindent If you are reading this inside of GNU Emacs, you can evaluate the function definition in the usual way to install it in Emacs, and then you can evaluate the following two expressions to see the results: @smallexample @group (type-of-animal "fierce") (type-of-animal "striped") @end group @end smallexample @c Following sentences rewritten to prevent overfull hbox. @noindent When you evaluate @code{(type-of-animal "fierce")}, you will see the following message printed in the echo area: @code{"It is a tiger!"}; and when you evaluate @code{(type-of-animal "striped")} you will see @code{nil} printed in the echo area. @node type-of-animal in detail @subsection The @code{type-of-animal} Function in Detail Let's look at the @code{type-of-animal} function in detail. The function definition for @code{type-of-animal} was written by filling the slots of two templates, one for a function definition as a whole, and a second for an @code{if} expression. @need 1250 The template for every function that is not interactive is: @smallexample @group (defun @var{name-of-function} (@var{argument-list}) "@var{documentation}@dots{}" @var{body}@dots{}) @end group @end smallexample @need 800 The parts of the function that match this template look like this: @smallexample @group (defun type-of-animal (characteristic) "Print message in echo area depending on CHARACTERISTIC. If the CHARACTERISTIC is the string \"fierce\", then warn of a tiger." @var{body: the} @code{if} @var{expression}) @end group @end smallexample The name of function is @code{type-of-animal}; it is passed the value of one argument. The argument list is followed by a multi-line documentation string. The documentation string is included in the example because it is a good habit to write documentation string for every function definition. The body of the function definition consists of the @code{if} expression. @need 800 The template for an @code{if} expression looks like this: @smallexample @group (if @var{true-or-false-test} @var{action-to-carry-out-if-the-test-returns-true}) @end group @end smallexample @need 1250 In the @code{type-of-animal} function, the code for the @code{if} looks like this: @smallexample @group (if (equal characteristic "fierce") (message "It is a tiger!"))) @end group @end smallexample @need 800 Here, the true-or-false-test is the expression: @smallexample (equal characteristic "fierce") @end smallexample @noindent In Lisp, @code{equal} is a function that determines whether its first argument is equal to its second argument. The second argument is the string @code{"fierce"} and the first argument is the value of the symbol @code{characteristic}---in other words, the argument passed to this function. In the first exercise of @code{type-of-animal}, the argument @code{"fierce"} is passed to @code{type-of-animal}. Since @code{"fierce"} is equal to @code{"fierce"}, the expression, @code{(equal characteristic "fierce")}, returns a value of true. When this happens, the @code{if} evaluates the second argument or then-part of the @code{if}: @code{(message "It is a tiger!")}. On the other hand, in the second exercise of @code{type-of-animal}, the argument @code{"striped"} is passed to @code{type-of-animal}. @code{"striped"} is not equal to @code{"fierce"}, so the then-part is not evaluated and @code{nil} is returned by the @code{if} expression. @node else @section If--then--else Expressions @cindex Else An @code{if} expression may have an optional third argument, called the @dfn{else-part}, for the case when the true-or-false-test returns false. When this happens, the second argument or then-part of the overall @code{if} expression is @emph{not} evaluated, but the third or else-part @emph{is} evaluated. You might think of this as the cloudy day alternative for the decision ``if it is warm and sunny, then go to the beach, else read a book!''. The word ``else'' is not written in the Lisp code; the else-part of an @code{if} expression comes after the then-part. In the written Lisp, the else-part is usually written to start on a line of its own and is indented less than the then-part: @smallexample @group (if @var{true-or-false-test} @var{action-to-carry-out-if-the-test-returns-true} @var{action-to-carry-out-if-the-test-returns-false}) @end group @end smallexample For example, the following @code{if} expression prints the message @samp{4 is not greater than 5!} when you evaluate it in the usual way: @smallexample @group (if (> 4 5) ; @r{if-part} (message "4 falsely greater than 5!") ; @r{then-part} (message "4 is not greater than 5!")) ; @r{else-part} @end group @end smallexample @noindent Note that the different levels of indentation make it easy to distinguish the then-part from the else-part. (GNU Emacs has several commands that automatically indent @code{if} expressions correctly. @xref{Typing Lists, , GNU Emacs Helps You Type Lists}.) We can extend the @code{type-of-animal} function to include an else-part by simply incorporating an additional part to the @code{if} expression. @need 1500 You can see the consequences of doing this if you evaluate the following version of the @code{type-of-animal} function definition to install it and then evaluate the two subsequent expressions to pass different arguments to the function. @smallexample @group (defun type-of-animal (characteristic) ; @r{Second version.} "Print message in echo area depending on CHARACTERISTIC. If the CHARACTERISTIC is the string \"fierce\", then warn of a tiger; else say it is not fierce." (if (equal characteristic "fierce") (message "It is a tiger!") (message "It is not fierce!"))) @end group @end smallexample @sp 1 @smallexample @group (type-of-animal "fierce") (type-of-animal "striped") @end group @end smallexample @c Following sentence rewritten to prevent overfull hbox. @noindent When you evaluate @code{(type-of-animal "fierce")}, you will see the following message printed in the echo area: @code{"It is a tiger!"}; but when you evaluate @code{(type-of-animal "striped")}, you will see @code{"It is not fierce!"}. (Of course, if the @var{characteristic} were @code{"ferocious"}, the message @code{"It is not fierce!"} would be printed; and it would be misleading! When you write code, you need to take into account the possibility that some such argument will be tested by the @code{if} and write your program accordingly.) @node Truth & Falsehood @section Truth and Falsehood in Emacs Lisp @cindex Truth and falsehood in Emacs Lisp @cindex Falsehood and truth in Emacs Lisp @findex nil There is an important aspect to the truth test in an @code{if} expression. So far, we have spoken of ``true'' and ``false'' as values of predicates as if they were new kinds of Emacs Lisp objects. In fact, ``false'' is just our old friend @code{nil}. Anything else---anything at all---is ``true''. The expression that tests for truth is interpreted as @dfn{true} if the result of evaluating it is a value that is not @code{nil}. In other words, the result of the test is considered true if the value returned is a number such as 47, a string such as @code{"hello"}, or a symbol (other than @code{nil}) such as @code{flowers}, or a list (so long as it is not empty), or even a buffer! @menu * nil explained:: @code{nil} has two meanings. @end menu @ifnottex @node nil explained @unnumberedsubsec An explanation of @code{nil} @end ifnottex Before illustrating a test for truth, we need an explanation of @code{nil}. In Emacs Lisp, the symbol @code{nil} has two meanings. First, it means the empty list. Second, it means false and is the value returned when a true-or-false-test tests false. @code{nil} can be written as an empty list, @code{()}, or as @code{nil}. As far as the Lisp interpreter is concerned, @code{()} and @code{nil} are the same. Humans, however, tend to use @code{nil} for false and @code{()} for the empty list. In Emacs Lisp, any value that is not @code{nil}---is not the empty list---is considered true. This means that if an evaluation returns something that is not an empty list, an @code{if} expression will test true. For example, if a number is put in the slot for the test, it will be evaluated and will return itself, since that is what numbers do when evaluated. In this conditional, the @code{if} expression will test true. The expression tests false only when @code{nil}, an empty list, is returned by evaluating the expression. You can see this by evaluating the two expressions in the following examples. In the first example, the number 4 is evaluated as the test in the @code{if} expression and returns itself; consequently, the then-part of the expression is evaluated and returned: @samp{true} appears in the echo area. In the second example, the @code{nil} indicates false; consequently, the else-part of the expression is evaluated and returned: @samp{false} appears in the echo area. @smallexample @group (if 4 'true 'false) @end group @group (if nil 'true 'false) @end group @end smallexample @need 1250 Incidentally, if some other useful value is not available for a test that returns true, then the Lisp interpreter will return the symbol @code{t} for true. For example, the expression @code{(> 5 4)} returns @code{t} when evaluated, as you can see by evaluating it in the usual way: @smallexample (> 5 4) @end smallexample @need 1250 @noindent On the other hand, this function returns @code{nil} if the test is false. @smallexample (> 4 5) @end smallexample @node save-excursion @section @code{save-excursion} @findex save-excursion @cindex Region, what it is @cindex Preserving point and buffer @cindex Point and buffer preservation @findex point @findex mark The @code{save-excursion} function is the final special form that we will discuss in this chapter. In Emacs Lisp programs used for editing, the @code{save-excursion} function is very common. It saves the location of point, executes the body of the function, and then restores point to its previous position if its location was changed. Its primary purpose is to keep the user from being surprised and disturbed by unexpected movement of point. @menu * Point and mark:: A review of various locations. * Template for save-excursion:: @end menu @ifnottex @node Point and mark @unnumberedsubsec Point and Mark @end ifnottex Before discussing @code{save-excursion}, however, it may be useful first to review what point and mark are in GNU Emacs. @dfn{Point} is the current location of the cursor. Wherever the cursor is, that is point. More precisely, on terminals where the cursor appears to be on top of a character, point is immediately before the character. In Emacs Lisp, point is an integer. The first character in a buffer is number one, the second is number two, and so on. The function @code{point} returns the current position of the cursor as a number. Each buffer has its own value for point. The @dfn{mark} is another position in the buffer; its value can be set with a command such as @kbd{C-@key{SPC}} (@code{set-mark-command}). If a mark has been set, you can use the command @kbd{C-x C-x} (@code{exchange-point-and-mark}) to cause the cursor to jump to the mark and set the mark to be the previous position of point. In addition, if you set another mark, the position of the previous mark is saved in the mark ring. Many mark positions can be saved this way. You can jump the cursor to a saved mark by typing @kbd{C-u C-@key{SPC}} one or more times. The part of the buffer between point and mark is called @dfn{the region}. Numerous commands work on the region, including @code{center-region}, @code{count-lines-region}, @code{kill-region}, and @code{print-region}. The @code{save-excursion} special form saves the location of point and restores this position after the code within the body of the special form is evaluated by the Lisp interpreter. Thus, if point were in the beginning of a piece of text and some code moved point to the end of the buffer, the @code{save-excursion} would put point back to where it was before, after the expressions in the body of the function were evaluated. In Emacs, a function frequently moves point as part of its internal workings even though a user would not expect this. For example, @code{count-lines-region} moves point. To prevent the user from being bothered by jumps that are both unexpected and (from the user's point of view) unnecessary, @code{save-excursion} is often used to keep point in the location expected by the user. The use of @code{save-excursion} is good housekeeping. To make sure the house stays clean, @code{save-excursion} restores the value of point even if something goes wrong in the code inside of it (or, to be more precise and to use the proper jargon, ``in case of abnormal exit''). This feature is very helpful. In addition to recording the value of point, @code{save-excursion} keeps track of the current buffer, and restores it, too. This means you can write code that will change the buffer and have @code{save-excursion} switch you back to the original buffer. This is how @code{save-excursion} is used in @code{append-to-buffer}. (@xref{append-to-buffer, , The Definition of @code{append-to-buffer}}.) @node Template for save-excursion @subsection Template for a @code{save-excursion} Expression @need 800 The template for code using @code{save-excursion} is simple: @smallexample @group (save-excursion @var{body}@dots{}) @end group @end smallexample @noindent The body of the function is one or more expressions that will be evaluated in sequence by the Lisp interpreter. If there is more than one expression in the body, the value of the last one will be returned as the value of the @code{save-excursion} function. The other expressions in the body are evaluated only for their side effects; and @code{save-excursion} itself is used only for its side effect (which is restoring the position of point). @need 1250 In more detail, the template for a @code{save-excursion} expression looks like this: @smallexample @group (save-excursion @var{first-expression-in-body} @var{second-expression-in-body} @var{third-expression-in-body} @dots{} @var{last-expression-in-body}) @end group @end smallexample @noindent An expression, of course, may be a symbol on its own or a list. In Emacs Lisp code, a @code{save-excursion} expression often occurs within the body of a @code{let} expression. It looks like this: @smallexample @group (let @var{varlist} (save-excursion @var{body}@dots{})) @end group @end smallexample @node Review @section Review In the last few chapters we have introduced a macro and a fair number of functions and special forms. Here they are described in brief, along with a few similar functions that have not been mentioned yet. @table @code @item eval-last-sexp Evaluate the last symbolic expression before the current location of point. The value is printed in the echo area unless the function is invoked with an argument; in that case, the output is printed in the current buffer. This command is normally bound to @kbd{C-x C-e}. @item defun Define function. This macro has up to five parts: the name, a template for the arguments that will be passed to the function, documentation, an optional interactive declaration, and the body of the definition. @need 1250 For example, in Emacs the function definition of @code{dired-unmark-all-marks} is as follows. @smallexample @group (defun dired-unmark-all-marks () "Remove all marks from all files in the Dired buffer." (interactive) (dired-unmark-all-files ?\r)) @end group @end smallexample @item interactive Declare to the interpreter that the function can be used interactively. This special form may be followed by a string with one or more parts that pass the information to the arguments of the function, in sequence. These parts may also tell the interpreter to prompt for information. Parts of the string are separated by newlines, @samp{\n}. @need 1000 Common code characters are: @table @code @item b The name of an existing buffer. @item f The name of an existing file. @item p The numeric prefix argument. (Note that this @code{p} is lower case.) @item r Point and the mark, as two numeric arguments, smallest first. This is the only code letter that specifies two successive arguments rather than one. @end table @xref{Interactive Codes, , Code Characters for @samp{interactive}, elisp, The GNU Emacs Lisp Reference Manual}, for a complete list of code characters. @item let Declare that a list of variables is for use within the body of the @code{let} and give them an initial value, either @code{nil} or a specified value; then evaluate the rest of the expressions in the body of the @code{let} and return the value of the last one. Inside the body of the @code{let}, the Lisp interpreter does not see the values of the variables of the same names that are bound outside of the @code{let}. @need 1250 For example, @smallexample @group (let ((foo (buffer-name)) (bar (buffer-size))) (message "This buffer is %s and has %d characters." foo bar)) @end group @end smallexample @item save-excursion Record the values of point and the current buffer before evaluating the body of this special form. Restore the value of point and buffer afterward. @need 1250 For example, @smallexample @group (message "We are %d characters into this buffer." (- (point) (save-excursion (goto-char (point-min)) (point)))) @end group @end smallexample @item if Evaluate the first argument to the function; if it is true, evaluate the second argument; else evaluate the third argument, if there is one. The @code{if} special form is called a @dfn{conditional}. There are other conditionals in Emacs Lisp, but @code{if} is perhaps the most commonly used. @need 1250 For example, @smallexample @group (if (= 22 emacs-major-version) (message "This is version 22 Emacs") (message "This is not version 22 Emacs")) @end group @end smallexample @need 1250 @item < @itemx > @itemx <= @itemx >= The @code{<} function tests whether its first argument is smaller than its second argument. A corresponding function, @code{>}, tests whether the first argument is greater than the second. Likewise, @code{<=} tests whether the first argument is less than or equal to the second and @code{>=} tests whether the first argument is greater than or equal to the second. In all cases, both arguments must be numbers or markers (markers indicate positions in buffers). @need 800 @item = The @code{=} function tests whether two arguments, both numbers or markers, are equal. @need 1250 @item equal @itemx eq Test whether two objects are the same. @code{equal} uses one meaning of the word ``same'' and @code{eq} uses another: @code{equal} returns true if the two objects have a similar structure and contents, such as two copies of the same book. On the other hand, @code{eq}, returns true if both arguments are actually the same object. @findex equal @findex eq @need 1250 @item string< @itemx string-lessp @itemx string= @itemx string-equal The @code{string-lessp} function tests whether its first argument is smaller than the second argument. A shorter, alternative name for the same function (a @code{defalias}) is @code{string<}. The arguments to @code{string-lessp} must be strings or symbols; the ordering is lexicographic, so case is significant. The print names of symbols are used instead of the symbols themselves. @cindex @samp{empty string} defined An empty string, @samp{""}, a string with no characters in it, is smaller than any string of characters. @code{string-equal} provides the corresponding test for equality. Its shorter, alternative name is @code{string=}. There are no string test functions that correspond to @var{>}, @code{>=}, or @code{<=}. @item message Print a message in the echo area. The first argument is a string that can contain @samp{%s}, @samp{%d}, or @samp{%c} to print the value of arguments that follow the string. The argument used by @samp{%s} must be a string or a symbol; the argument used by @samp{%d} must be a number. The argument used by @samp{%c} must be an @sc{ascii} code number; it will be printed as the character with that @sc{ascii} code. (Various other %-sequences have not been mentioned.) @item setq @itemx set The @code{setq} special form sets the value of its first argument to the value of the second argument. The first argument is automatically quoted by @code{setq}. It does the same for succeeding pairs of arguments. Another function, @code{set}, takes only two arguments and evaluates both of them before setting the value returned by its first argument to the value returned by its second argument. @item buffer-name Without an argument, return the name of the buffer, as a string. @item buffer-file-name Without an argument, return the name of the file the buffer is visiting. @item current-buffer Return the buffer in which Emacs is active; it may not be the buffer that is visible on the screen. @item other-buffer Return the most recently selected buffer (other than the buffer passed to @code{other-buffer} as an argument and other than the current buffer). @item switch-to-buffer Select a buffer for Emacs to be active in and display it in the current window so users can look at it. Usually bound to @kbd{C-x b}. @item set-buffer Switch Emacs's attention to a buffer on which programs will run. Don't alter what the window is showing. @item buffer-size Return the number of characters in the current buffer. @item point Return the value of the current position of the cursor, as an integer counting the number of characters from the beginning of the buffer. @item point-min Return the minimum permissible value of point in the current buffer. This is 1, unless narrowing is in effect. @item point-max Return the value of the maximum permissible value of point in the current buffer. This is the end of the buffer, unless narrowing is in effect. @end table @need 1500 @node defun Exercises @section Exercises @itemize @bullet @item Write a non-interactive function that doubles the value of its argument, a number. Make that function interactive. @item Write a function that tests whether the current value of @code{fill-column} is greater than the argument passed to the function, and if so, prints an appropriate message. @end itemize @node Buffer Walk Through @chapter A Few Buffer-Related Functions In this chapter we study in detail several of the functions used in GNU Emacs. This is called a ``walk-through''. These functions are used as examples of Lisp code, but are not imaginary examples; with the exception of the first, simplified function definition, these functions show the actual code used in GNU Emacs. You can learn a great deal from these definitions. The functions described here are all related to buffers. Later, we will study other functions. @menu * Finding More:: How to find more information. * simplified-beginning-of-buffer:: Shows @code{goto-char}, @code{point-min}, and @code{push-mark}. * mark-whole-buffer:: Almost the same as @code{beginning-of-buffer}. * append-to-buffer:: Uses @code{save-excursion} and @code{insert-buffer-substring}. * Buffer Related Review:: Review. * Buffer Exercises:: @end menu @node Finding More @section Finding More Information @findex describe-function@r{, introduced} @cindex Find function documentation In this walk-through, I will describe each new function as we come to it, sometimes in detail and sometimes briefly. If you are interested, you can get the full documentation of any Emacs Lisp function at any time by typing @kbd{C-h f} and then the name of the function (and then @key{RET}). Similarly, you can get the full documentation for a variable by typing @kbd{C-h v} and then the name of the variable (and then @key{RET}). @cindex Find source of function @c In version 22, tells location both of C and of Emacs Lisp Also, @code{describe-function} will tell you the location of the function definition. Put point into the name of the file that contains the function and press the @key{RET} key. In this case, @key{RET} means @code{push-button} rather than ``return'' or ``enter''. Emacs will take you directly to the function definition. @ignore Not In version 22 If you move point over the file name and press the @key{RET} key, which in this case means @code{help-follow} rather than ``return'' or ``enter'', Emacs will take you directly to the function definition. @end ignore More generally, if you want to see a function in its original source file, you can use the @code{xref-find-definitions} function to jump to it. @code{xref-find-definitions} works with a wide variety of languages, not just Lisp, and C, and it works with non-programming text as well. For example, @code{xref-find-definitions} will jump to the various nodes in the Texinfo source file of this document (provided that you've run the @command{etags} utility to record all the nodes in the manuals that come with Emacs; @pxref{Create Tags Table,,, emacs, The GNU Emacs Manual}). To use the @code{xref-find-definitions} command, type @kbd{M-.} (i.e., press the period key while holding down the @key{META} key, or else type the @key{ESC} key and then type the period key), and then, at the prompt, type in the name of the function whose source code you want to see, such as @code{mark-whole-buffer}, and then type @key{RET}. (If the command doesn't prompt, invoke it with an argument: @kbd{C-u M-.}; @pxref{Interactive Options}.) Emacs will switch buffers and display the source code for the function on your screen@footnote{ If instead of showing the source code for a Lisp function, Emacs asks you which tags table to visit, invoke @kbd{M-.} from a buffer whose major mode is Emacs Lisp or Lisp Interaction. }. To switch back to your current buffer, type @kbd{M-,} or @kbd{C-x b @key{RET}}. (On some keyboards, the @key{META} key is labeled @key{ALT}.) @cindex Library, as term for ``file'' Incidentally, the files that contain Lisp code are conventionally called @dfn{libraries}. The metaphor is derived from that of a specialized library, such as a law library or an engineering library, rather than a general library. Each library, or file, contains functions that relate to a particular topic or activity, such as @file{abbrev.el} for handling abbreviations and other typing shortcuts, and @file{help.el} for help. (Sometimes several libraries provide code for a single activity, as the various @file{rmail@dots{}} files provide code for reading electronic mail.) In @cite{The GNU Emacs Manual}, you will see sentences such as ``The @kbd{C-h p} command lets you search the standard Emacs Lisp libraries by topic keywords.'' @node simplified-beginning-of-buffer @section A Simplified @code{beginning-of-buffer} Definition @findex simplified-beginning-of-buffer The @code{beginning-of-buffer} command is a good function to start with since you are likely to be familiar with it and it is easy to understand. Used as an interactive command, @code{beginning-of-buffer} moves the cursor to the beginning of the buffer, leaving the mark at the previous position. It is generally bound to @kbd{M-<}. In this section, we will discuss a shortened version of the function that shows how it is most frequently used. This shortened function works as written, but it does not contain the code for a complex option. In another section, we will describe the entire function. (@xref{beginning-of-buffer, , Complete Definition of @code{beginning-of-buffer}}.) Before looking at the code, let's consider what the function definition has to contain: it must include an expression that makes the function interactive so it can be called by typing @kbd{M-x beginning-of-buffer} or by typing a keychord such as @kbd{M-<}; it must include code to leave a mark at the original position in the buffer; and it must include code to move the cursor to the beginning of the buffer. @need 1250 Here is the complete text of the shortened version of the function: @smallexample @group (defun simplified-beginning-of-buffer () "Move point to the beginning of the buffer; leave mark at previous position." (interactive) (push-mark) (goto-char (point-min))) @end group @end smallexample Like all function definitions, this definition has five parts following the macro @code{defun}: @enumerate @item The name: in this example, @code{simplified-beginning-of-buffer}. @item A list of the arguments: in this example, an empty list, @code{()}, @item The documentation string. @item The interactive expression. @item The body. @end enumerate @noindent In this function definition, the argument list is empty; this means that this function does not require any arguments. (When we look at the definition for the complete function, we will see that it may be passed an optional argument.) The interactive expression tells Emacs that the function is intended to be used interactively. In this example, @code{interactive} does not have an argument because @code{simplified-beginning-of-buffer} does not require one. @need 800 The body of the function consists of the two lines: @smallexample @group (push-mark) (goto-char (point-min)) @end group @end smallexample The first of these lines is the expression, @code{(push-mark)}. When this expression is evaluated by the Lisp interpreter, it sets a mark at the current position of the cursor, wherever that may be. The position of this mark is saved in the mark ring. The next line is @code{(goto-char (point-min))}. This expression jumps the cursor to the minimum point in the buffer, that is, to the beginning of the buffer (or to the beginning of the accessible portion of the buffer if it is narrowed. @xref{Narrowing & Widening, , Narrowing and Widening}.) The @code{push-mark} command sets a mark at the place where the cursor was located before it was moved to the beginning of the buffer by the @code{(goto-char (point-min))} expression. Consequently, you can, if you wish, go back to where you were originally by typing @kbd{C-x C-x}. That is all there is to the function definition! @findex describe-function When you are reading code such as this and come upon an unfamiliar function, such as @code{goto-char}, you can find out what it does by using the @code{describe-function} command. To use this command, type @kbd{C-h f} and then type in the name of the function and press @key{RET}. The @code{describe-function} command will print the function's documentation string in a @file{*Help*} window. For example, the documentation for @code{goto-char} is: @smallexample @group Set point to POSITION, a number or marker. Beginning of buffer is position (point-min), end is (point-max). @end group @end smallexample @noindent The function's one argument is the desired position. @noindent (The prompt for @code{describe-function} will offer you the symbol under or preceding the cursor, so you can save typing by positioning the cursor right over or after the function and then typing @kbd{C-h f @key{RET}}.) The @code{end-of-buffer} function definition is written in the same way as the @code{beginning-of-buffer} definition except that the body of the function contains the expression @code{(goto-char (point-max))} in place of @code{(goto-char (point-min))}. @node mark-whole-buffer @section The Definition of @code{mark-whole-buffer} @findex mark-whole-buffer The @code{mark-whole-buffer} function is no harder to understand than the @code{simplified-beginning-of-buffer} function. In this case, however, we will look at the complete function, not a shortened version. The @code{mark-whole-buffer} function is not as commonly used as the @code{beginning-of-buffer} function, but is useful nonetheless: it marks a whole buffer as a region by putting point at the beginning and a mark at the end of the buffer. It is generally bound to @kbd{C-x h}. @menu * mark-whole-buffer overview:: * Body of mark-whole-buffer:: Only three lines of code. @end menu @ifnottex @node mark-whole-buffer overview @unnumberedsubsec An overview of @code{mark-whole-buffer} @end ifnottex @need 1250 In GNU Emacs 22, the code for the complete function looks like this: @smallexample @group (defun mark-whole-buffer () "Put point at beginning and mark at end of buffer. You probably should not use this function in Lisp programs; it is usually a mistake for a Lisp function to use any subroutine that uses or sets the mark." (interactive) (push-mark (point)) (push-mark (point-max) nil t) (goto-char (point-min))) @end group @end smallexample @need 1250 Like all other functions, the @code{mark-whole-buffer} function fits into the template for a function definition. The template looks like this: @smallexample @group (defun @var{name-of-function} (@var{argument-list}) "@var{documentation}@dots{}" (@var{interactive-expression}@dots{}) @var{body}@dots{}) @end group @end smallexample Here is how the function works: the name of the function is @code{mark-whole-buffer}; it is followed by an empty argument list, @samp{()}, which means that the function does not require arguments. The documentation comes next. The next line is an @code{(interactive)} expression that tells Emacs that the function will be used interactively. These details are similar to the @code{simplified-beginning-of-buffer} function described in the previous section. @need 1250 @node Body of mark-whole-buffer @subsection Body of @code{mark-whole-buffer} The body of the @code{mark-whole-buffer} function consists of three lines of code: @c GNU Emacs 22 @smallexample @group (push-mark (point)) (push-mark (point-max) nil t) (goto-char (point-min)) @end group @end smallexample The first of these lines is the expression, @code{(push-mark (point))}. This line does exactly the same job as the first line of the body of the @code{simplified-beginning-of-buffer} function, which is written @code{(push-mark)}. In both cases, the Lisp interpreter sets a mark at the current position of the cursor. I don't know why the expression in @code{mark-whole-buffer} is written @code{(push-mark (point))} and the expression in @code{beginning-of-buffer} is written @code{(push-mark)}. Perhaps whoever wrote the code did not know that the arguments for @code{push-mark} are optional and that if @code{push-mark} is not passed an argument, the function automatically sets mark at the location of point by default. Or perhaps the expression was written so as to parallel the structure of the next line. In any case, the line causes Emacs to determine the position of point and set a mark there. In earlier versions of GNU Emacs, the next line of @code{mark-whole-buffer} was @code{(push-mark (point-max))}. This expression sets a mark at the point in the buffer that has the highest number. This will be the end of the buffer (or, if the buffer is narrowed, the end of the accessible portion of the buffer. @xref{Narrowing & Widening, , Narrowing and Widening}, for more about narrowing.) After this mark has been set, the previous mark, the one set at point, is no longer set, but Emacs remembers its position, just as all other recent marks are always remembered. This means that you can, if you wish, go back to that position by typing @kbd{C-u C-@key{SPC}} twice. @need 1250 In GNU Emacs 22, the @code{(point-max)} is slightly more complicated. The line reads @smallexample (push-mark (point-max) nil t) @end smallexample @noindent The expression works nearly the same as before. It sets a mark at the highest numbered place in the buffer that it can. However, in this version, @code{push-mark} has two additional arguments. The second argument to @code{push-mark} is @code{nil}. This tells the function it @emph{should} display a message that says ``Mark set'' when it pushes the mark. The third argument is @code{t}. This tells @code{push-mark} to activate the mark when Transient Mark mode is turned on. Transient Mark mode highlights the currently active region. It is often turned off. Finally, the last line of the function is @code{(goto-char (point-min)))}. This is written exactly the same way as it is written in @code{beginning-of-buffer}. The expression moves the cursor to the minimum point in the buffer, that is, to the beginning of the buffer (or to the beginning of the accessible portion of the buffer). As a result of this, point is placed at the beginning of the buffer and mark is set at the end of the buffer. The whole buffer is, therefore, the region. @c FIXME: the definition of append-to-buffer has been changed (in @c 2010-03-30). @node append-to-buffer @section The Definition of @code{append-to-buffer} @findex append-to-buffer The @code{append-to-buffer} command is more complex than the @code{mark-whole-buffer} command. What it does is copy the region (that is, the part of the buffer between point and mark) from the current buffer to a specified buffer. @menu * append-to-buffer overview:: * append interactive:: A two part interactive expression. * append-to-buffer body:: Incorporates a @code{let} expression. * append save-excursion:: How the @code{save-excursion} works. @end menu @ifnottex @node append-to-buffer overview @unnumberedsubsec An Overview of @code{append-to-buffer} @end ifnottex @findex insert-buffer-substring The @code{append-to-buffer} command uses the @code{insert-buffer-substring} function to copy the region. @code{insert-buffer-substring} is described by its name: it takes a substring from a buffer, and inserts it into another buffer. Most of @code{append-to-buffer} is concerned with setting up the conditions for @code{insert-buffer-substring} to work: the code must specify both the buffer to which the text will go, the window it comes from and goes to, and the region that will be copied. @need 1250 Here is the complete text of the function: @smallexample @group (defun append-to-buffer (buffer start end) "Append to specified buffer the text of the region. It is inserted into that buffer before its point. @end group @group When calling from a program, give three arguments: BUFFER (or buffer name), START and END. START and END specify the portion of the current buffer to be copied." (interactive (list (read-buffer "Append to buffer: " (other-buffer (current-buffer) t)) (region-beginning) (region-end))) @end group @group (let ((oldbuf (current-buffer))) (save-excursion (let* ((append-to (get-buffer-create buffer)) (windows (get-buffer-window-list append-to t t)) point) (set-buffer append-to) (setq point (point)) (barf-if-buffer-read-only) (insert-buffer-substring oldbuf start end) (dolist (window windows) (when (= (window-point window) point) (set-window-point window (point)))))))) @end group @end smallexample The function can be understood by looking at it as a series of filled-in templates. The outermost template is for the function definition. In this function, it looks like this (with several slots filled in): @smallexample @group (defun append-to-buffer (buffer start end) "@var{documentation}@dots{}" (interactive @dots{}) @var{body}@dots{}) @end group @end smallexample The first line of the function includes its name and three arguments. The arguments are the @code{buffer} to which the text will be copied, and the @code{start} and @code{end} of the region in the current buffer that will be copied. The next part of the function is the documentation, which is clear and complete. As is conventional, the three arguments are written in upper case so you will notice them easily. Even better, they are described in the same order as in the argument list. Note that the documentation distinguishes between a buffer and its name. (The function can handle either.) @node append interactive @subsection The @code{append-to-buffer} Interactive Expression Since the @code{append-to-buffer} function will be used interactively, the function must have an @code{interactive} expression. (For a review of @code{interactive}, see @ref{Interactive, , Making a Function Interactive}.) The expression reads as follows: @smallexample @group (interactive (list (read-buffer "Append to buffer: " (other-buffer (current-buffer) t)) (region-beginning) (region-end))) @end group @end smallexample @noindent This expression is not one with letters standing for parts, as described earlier. Instead, it starts a list with these parts: The first part of the list is an expression to read the name of a buffer and return it as a string. That is @code{read-buffer}. The function requires a prompt as its first argument, @samp{"Append to buffer: "}. Its second argument tells the command what value to provide if you don't specify anything. In this case that second argument is an expression containing the function @code{other-buffer}, an exception, and a @samp{t}, standing for true. The first argument to @code{other-buffer}, the exception, is yet another function, @code{current-buffer}. That is not going to be returned. The second argument is the symbol for true, @code{t}. that tells @code{other-buffer} that it may show visible buffers (except in this case, it will not show the current buffer, which makes sense). @need 1250 The expression looks like this: @smallexample (other-buffer (current-buffer) t) @end smallexample The second and third arguments to the @code{list} expression are @code{(region-beginning)} and @code{(region-end)}. These two functions specify the beginning and end of the text to be appended. @need 1250 Originally, the command used the letters @samp{B} and @samp{r}. The whole @code{interactive} expression looked like this: @smallexample (interactive "BAppend to buffer:@: \nr") @end smallexample @noindent But when that was done, the default value of the buffer switched to was invisible. That was not wanted. (The prompt was separated from the second argument with a newline, @samp{\n}. It was followed by an @samp{r} that told Emacs to bind the two arguments that follow the symbol @code{buffer} in the function's argument list (that is, @code{start} and @code{end}) to the values of point and mark. That argument worked fine.) @node append-to-buffer body @subsection The Body of @code{append-to-buffer} @ignore in GNU Emacs 22 in /usr/local/src/emacs/lisp/simple.el (defun append-to-buffer (buffer start end) "Append to specified buffer the text of the region. It is inserted into that buffer before its point. When calling from a program, give three arguments: BUFFER (or buffer name), START and END. START and END specify the portion of the current buffer to be copied." (interactive (list (read-buffer "Append to buffer: " (other-buffer (current-buffer) t)) (region-beginning) (region-end))) (let ((oldbuf (current-buffer))) (save-excursion (let* ((append-to (get-buffer-create buffer)) (windows (get-buffer-window-list append-to t t)) point) (set-buffer append-to) (setq point (point)) (barf-if-buffer-read-only) (insert-buffer-substring oldbuf start end) (dolist (window windows) (when (= (window-point window) point) (set-window-point window (point)))))))) @end ignore The body of the @code{append-to-buffer} function begins with @code{let}. As we have seen before (@pxref{let, , @code{let}}), the purpose of a @code{let} expression is to create and give initial values to one or more variables that will only be used within the body of the @code{let}. This means that such a variable will not be confused with any variable of the same name outside the @code{let} expression. We can see how the @code{let} expression fits into the function as a whole by showing a template for @code{append-to-buffer} with the @code{let} expression in outline: @smallexample @group (defun append-to-buffer (buffer start end) "@var{documentation}@dots{}" (interactive @dots{}) (let ((@var{variable} @var{value})) @var{body}@dots{}) @end group @end smallexample The @code{let} expression has three elements: @enumerate @item The symbol @code{let}; @item A varlist containing, in this case, a single two-element list, @code{(@var{variable} @var{value})}; @item The body of the @code{let} expression. @end enumerate @need 800 In the @code{append-to-buffer} function, the varlist looks like this: @smallexample (oldbuf (current-buffer)) @end smallexample @noindent In this part of the @code{let} expression, the one variable, @code{oldbuf}, is bound to the value returned by the @code{(current-buffer)} expression. The variable, @code{oldbuf}, is used to keep track of the buffer in which you are working and from which you will copy. The element or elements of a varlist are surrounded by a set of parentheses so the Lisp interpreter can distinguish the varlist from the body of the @code{let}. As a consequence, the two-element list within the varlist is surrounded by a circumscribing set of parentheses. The line looks like this: @smallexample @group (let ((oldbuf (current-buffer))) @dots{} ) @end group @end smallexample @noindent The two parentheses before @code{oldbuf} might surprise you if you did not realize that the first parenthesis before @code{oldbuf} marks the boundary of the varlist and the second parenthesis marks the beginning of the two-element list, @code{(oldbuf (current-buffer))}. @node append save-excursion @subsection @code{save-excursion} in @code{append-to-buffer} The body of the @code{let} expression in @code{append-to-buffer} consists of a @code{save-excursion} expression. The @code{save-excursion} function saves the location of point, and restores it to that position after the expressions in the body of the @code{save-excursion} complete execution. In addition, @code{save-excursion} keeps track of the original buffer, and restores it. This is how @code{save-excursion} is used in @code{append-to-buffer}. @need 1500 @cindex Indentation for formatting @cindex Formatting convention Incidentally, it is worth noting here that a Lisp function is normally formatted so that everything that is enclosed in a multi-line spread is indented more to the right than the first symbol. In this function definition, the @code{let} is indented more than the @code{defun}, and the @code{save-excursion} is indented more than the @code{let}, like this: @smallexample @group (defun @dots{} @dots{} @dots{} (let@dots{} (save-excursion @dots{} @end group @end smallexample @need 1500 @noindent This formatting convention makes it easy to see that the lines in the body of the @code{save-excursion} are enclosed by the parentheses associated with @code{save-excursion}, just as the @code{save-excursion} itself is enclosed by the parentheses associated with the @code{let}: @smallexample @group (let ((oldbuf (current-buffer))) (save-excursion @dots{} (set-buffer @dots{}) (insert-buffer-substring oldbuf start end) @dots{})) @end group @end smallexample @need 1200 The use of the @code{save-excursion} function can be viewed as a process of filling in the slots of a template: @smallexample @group (save-excursion @var{first-expression-in-body} @var{second-expression-in-body} @dots{} @var{last-expression-in-body}) @end group @end smallexample @need 1200 @noindent In this function, the body of the @code{save-excursion} contains only one expression, the @code{let*} expression. You know about a @code{let} function. The @code{let*} function is different. It has a @samp{*} in its name. It enables Emacs to set each variable in its varlist in sequence, one after another. Its critical feature is that variables later in the varlist can make use of the values to which Emacs set variables earlier in the varlist. @xref{fwd-para let, , The @code{let*} expression}. We will skip functions like @code{let*} and focus on two: the @code{set-buffer} function and the @code{insert-buffer-substring} function. @need 1250 In the old days, the @code{set-buffer} expression was simply @smallexample (set-buffer (get-buffer-create buffer)) @end smallexample @need 1250 @noindent but now it is @smallexample (set-buffer append-to) @end smallexample @noindent @code{append-to} is bound to @code{(get-buffer-create buffer)} earlier on in the @code{let*} expression. That extra binding would not be necessary except for that @code{append-to} is used later in the varlist as an argument to @code{get-buffer-window-list}. @ignore in GNU Emacs 22 (let ((oldbuf (current-buffer))) (save-excursion (let* ((append-to (get-buffer-create buffer)) (windows (get-buffer-window-list append-to t t)) point) (set-buffer append-to) (setq point (point)) (barf-if-buffer-read-only) (insert-buffer-substring oldbuf start end) (dolist (window windows) (when (= (window-point window) point) (set-window-point window (point)))))))) @end ignore The @code{append-to-buffer} function definition inserts text from the buffer in which you are currently to a named buffer. It happens that @code{insert-buffer-substring} does just the reverse---it copies text from another buffer to the current buffer---that is why the @code{append-to-buffer} definition starts out with a @code{let} that binds the local symbol @code{oldbuf} to the value returned by @code{current-buffer}. @need 1250 The @code{insert-buffer-substring} expression looks like this: @smallexample (insert-buffer-substring oldbuf start end) @end smallexample @noindent The @code{insert-buffer-substring} function copies a string @emph{from} the buffer specified as its first argument and inserts the string into the present buffer. In this case, the argument to @code{insert-buffer-substring} is the value of the variable created and bound by the @code{let}, namely the value of @code{oldbuf}, which was the current buffer when you gave the @code{append-to-buffer} command. After @code{insert-buffer-substring} has done its work, @code{save-excursion} will restore the action to the original buffer and @code{append-to-buffer} will have done its job. @need 800 Written in skeletal form, the workings of the body look like this: @smallexample @group (let (@var{bind-}@code{oldbuf}@var{-to-value-of-}@code{current-buffer}) (save-excursion ; @r{Keep track of buffer.} @var{change-buffer} @var{insert-substring-from-}@code{oldbuf}@var{-into-buffer}) @var{change-back-to-original-buffer-when-finished} @var{let-the-local-meaning-of-}@code{oldbuf}@var{-disappear-when-finished} @end group @end smallexample In summary, @code{append-to-buffer} works as follows: it saves the value of the current buffer in the variable called @code{oldbuf}. It gets the new buffer (creating one if need be) and switches Emacs's attention to it. Using the value of @code{oldbuf}, it inserts the region of text from the old buffer into the new buffer; and then using @code{save-excursion}, it brings you back to your original buffer. In looking at @code{append-to-buffer}, you have explored a fairly complex function. It shows how to use @code{let} and @code{save-excursion}, and how to change to and come back from another buffer. Many function definitions use @code{let}, @code{save-excursion}, and @code{set-buffer} this way. @node Buffer Related Review @section Review Here is a brief summary of the various functions discussed in this chapter. @table @code @item describe-function @itemx describe-variable Print the documentation for a function or variable. Conventionally bound to @kbd{C-h f} and @kbd{C-h v}. @item xref-find-definitions Find the file containing the source for a function or variable and switch buffers to it, positioning point at the beginning of the item. Conventionally bound to @kbd{M-.} (that's a period following the @key{META} key). @item save-excursion Save the location of point and restore its value after the arguments to @code{save-excursion} have been evaluated. Also, remember the current buffer and return to it. @item push-mark Set mark at a location and record the value of the previous mark on the mark ring. The mark is a location in the buffer that will keep its relative position even if text is added to or removed from the buffer. @item goto-char Set point to the location specified by the value of the argument, which can be a number, a marker, or an expression that returns the number of a position, such as @code{(point-min)}. @item insert-buffer-substring Copy a region of text from a buffer that is passed to the function as an argument and insert the region into the current buffer. @item mark-whole-buffer Mark the whole buffer as a region. Normally bound to @kbd{C-x h}. @item set-buffer Switch the attention of Emacs to another buffer, but do not change the window being displayed. Used when the program rather than a human is to work on a different buffer. @item get-buffer-create @itemx get-buffer Find a named buffer or create one if a buffer of that name does not exist. The @code{get-buffer} function returns @code{nil} if the named buffer does not exist. @end table @need 1500 @node Buffer Exercises @section Exercises @itemize @bullet @item Write your own @code{simplified-end-of-buffer} function definition; then test it to see whether it works. @item Use @code{if} and @code{get-buffer} to write a function that prints a message telling you whether a buffer exists. @item Using @code{xref-find-definitions}, find the source for the @code{copy-to-buffer} function. @end itemize @node More Complex @chapter A Few More Complex Functions In this chapter, we build on what we have learned in previous chapters by looking at more complex functions. The @code{copy-to-buffer} function illustrates use of two @code{save-excursion} expressions in one definition, while the @code{insert-buffer} function illustrates use of an asterisk in an @code{interactive} expression, use of @code{or}, and the important distinction between a name and the object to which the name refers. @menu * copy-to-buffer:: With @code{set-buffer}, @code{get-buffer-create}. * insert-buffer:: Read-only, and with @code{or}. * beginning-of-buffer:: Shows @code{goto-char}, @code{point-min}, and @code{push-mark}. * Second Buffer Related Review:: * optional Exercise:: @end menu @node copy-to-buffer @section The Definition of @code{copy-to-buffer} @findex copy-to-buffer After understanding how @code{append-to-buffer} works, it is easy to understand @code{copy-to-buffer}. This function copies text into a buffer, but instead of adding to the second buffer, it replaces all the previous text in the second buffer. @need 800 The body of @code{copy-to-buffer} looks like this, @smallexample @group @dots{} (interactive "BCopy to buffer: \nr") (let ((oldbuf (current-buffer))) (with-current-buffer (get-buffer-create buffer) (barf-if-buffer-read-only) (erase-buffer) (save-excursion (insert-buffer-substring oldbuf start end))))) @end group @end smallexample The @code{copy-to-buffer} function has a simpler @code{interactive} expression than @code{append-to-buffer}. @need 800 The definition then says @smallexample (with-current-buffer (get-buffer-create buffer) @dots{} @end smallexample First, look at the earliest inner expression; that is evaluated first. That expression starts with @code{get-buffer-create buffer}. The function tells the computer to use the buffer with the name specified as the one to which you are copying, or if such a buffer does not exist, to create it. Then, the @code{with-current-buffer} function evaluates its body with that buffer temporarily current. (This demonstrates another way to shift the computer's attention but not the user's. The @code{append-to-buffer} function showed how to do the same with @code{save-excursion} and @code{set-buffer}. @code{with-current-buffer} is a newer, and arguably easier, mechanism.) The @code{barf-if-buffer-read-only} function sends you an error message saying the buffer is read-only if you cannot modify it. The next line has the @code{erase-buffer} function as its sole contents. That function erases the buffer. Finally, the last two lines contain the @code{save-excursion} expression with @code{insert-buffer-substring} as its body. The @code{insert-buffer-substring} expression copies the text from the buffer you are in (and you have not seen the computer shift its attention, so you don't know that that buffer is now called @code{oldbuf}). Incidentally, this is what is meant by ``replacement''. To replace text, Emacs erases the previous text and then inserts new text. @need 1250 In outline, the body of @code{copy-to-buffer} looks like this: @smallexample @group (let (@var{bind-}@code{oldbuf}@var{-to-value-of-}@code{current-buffer}) (@var{with-the-buffer-you-are-copying-to} (@var{but-do-not-erase-or-copy-to-a-read-only-buffer}) (erase-buffer) (save-excursion @var{insert-substring-from-}@code{oldbuf}@var{-into-buffer}))) @end group @end smallexample @node insert-buffer @section The Definition of @code{insert-buffer} @findex insert-buffer @code{insert-buffer} is yet another buffer-related function. This command copies another buffer @emph{into} the current buffer. It is the reverse of @code{append-to-buffer} or @code{copy-to-buffer}, since they copy a region of text @emph{from} the current buffer to another buffer. Here is a discussion based on the original code. The code was simplified in 2003 and is harder to understand. (@xref{New insert-buffer, , New Body for @code{insert-buffer}}, to see a discussion of the new body.) In addition, this code illustrates the use of @code{interactive} with a buffer that might be @dfn{read-only} and the important distinction between the name of an object and the object actually referred to. @menu * insert-buffer code:: * insert-buffer interactive:: When you can read, but not write. * insert-buffer body:: The body has an @code{or} and a @code{let}. * if & or:: Using an @code{if} instead of an @code{or}. * Insert or:: How the @code{or} expression works. * Insert let:: Two @code{save-excursion} expressions. * New insert-buffer:: @end menu @ifnottex @node insert-buffer code @unnumberedsubsec The Code for @code{insert-buffer} @end ifnottex @need 800 Here is the earlier code: @smallexample @group (defun insert-buffer (buffer) "Insert after point the contents of BUFFER. Puts mark after the inserted text. BUFFER may be a buffer or a buffer name." (interactive "*bInsert buffer:@: ") @end group @group (or (bufferp buffer) (setq buffer (get-buffer buffer))) (let (start end newmark) (save-excursion (save-excursion (set-buffer buffer) (setq start (point-min) end (point-max))) @end group @group (insert-buffer-substring buffer start end) (setq newmark (point))) (push-mark newmark))) @end group @end smallexample @need 1200 As with other function definitions, you can use a template to see an outline of the function: @smallexample @group (defun insert-buffer (buffer) "@var{documentation}@dots{}" (interactive "*bInsert buffer:@: ") @var{body}@dots{}) @end group @end smallexample @node insert-buffer interactive @subsection The Interactive Expression in @code{insert-buffer} @findex interactive@r{, example use of} In @code{insert-buffer}, the argument to the @code{interactive} declaration has two parts, an asterisk, @samp{*}, and @samp{bInsert buffer:@: }. @menu * Read-only buffer:: When a buffer cannot be modified. * b for interactive:: An existing buffer or else its name. @end menu @node Read-only buffer @unnumberedsubsubsec A Read-only Buffer @cindex Read-only buffer @cindex Asterisk for read-only buffer @findex * @r{for read-only buffer} The asterisk is for the situation when the current buffer is a read-only buffer---a buffer that cannot be modified. If @code{insert-buffer} is called when the current buffer is read-only, a message to this effect is printed in the echo area and the terminal may beep or blink at you; you will not be permitted to insert anything into current buffer. The asterisk does not need to be followed by a newline to separate it from the next argument. @node b for interactive @unnumberedsubsubsec @samp{b} in an Interactive Expression The next argument in the interactive expression starts with a lower case @samp{b}. (This is different from the code for @code{append-to-buffer}, which uses an upper-case @samp{B}. @xref{append-to-buffer, , The Definition of @code{append-to-buffer}}.) The lower-case @samp{b} tells the Lisp interpreter that the argument for @code{insert-buffer} should be an existing buffer or else its name. (The upper-case @samp{B} option provides for the possibility that the buffer does not exist.) Emacs will prompt you for the name of the buffer, offering you a default buffer, with name completion enabled. If the buffer does not exist, you receive a message that says ``No match''; your terminal may beep at you as well. The new and simplified code generates a list for @code{interactive}. It uses the @code{barf-if-buffer-read-only} and @code{read-buffer} functions with which we are already familiar and the @code{progn} special form with which we are not. (It will be described later.) @node insert-buffer body @subsection The Body of the @code{insert-buffer} Function The body of the @code{insert-buffer} function has two major parts: an @code{or} expression and a @code{let} expression. The purpose of the @code{or} expression is to ensure that the argument @code{buffer} is bound to a buffer and not just the name of a buffer. The body of the @code{let} expression contains the code which copies the other buffer into the current buffer. @need 1250 In outline, the two expressions fit into the @code{insert-buffer} function like this: @smallexample @group (defun insert-buffer (buffer) "@var{documentation}@dots{}" (interactive "*bInsert buffer:@: ") (or @dots{} @dots{} @end group @group (let (@var{varlist}) @var{body-of-}@code{let}@dots{} ) @end group @end smallexample To understand how the @code{or} expression ensures that the argument @code{buffer} is bound to a buffer and not to the name of a buffer, it is first necessary to understand the @code{or} function. Before doing this, let me rewrite this part of the function using @code{if} so that you can see what is done in a manner that will be familiar. @node if & or @subsection @code{insert-buffer} With an @code{if} Instead of an @code{or} The job to be done is to make sure the value of @code{buffer} is a buffer itself and not the name of a buffer. If the value is the name, then the buffer itself must be got. You can imagine yourself at a conference where an usher is wandering around holding a list with your name on it and looking for you: the usher is bound to your name, not to you; but when the usher finds you and takes your arm, the usher becomes bound to you. @need 800 In Lisp, you might describe this situation like this: @smallexample @group (if (not (holding-on-to-guest)) (find-and-take-arm-of-guest)) @end group @end smallexample We want to do the same thing with a buffer---if we do not have the buffer itself, we want to get it. @need 1200 Using a predicate called @code{bufferp} that tells us whether we have a buffer (rather than its name), we can write the code like this: @smallexample @group (if (not (bufferp buffer)) ; @r{if-part} (setq buffer (get-buffer buffer))) ; @r{then-part} @end group @end smallexample @noindent Here, the true-or-false-test of the @code{if} expression is @w{@code{(not (bufferp buffer))}}; and the then-part is the expression @w{@code{(setq buffer (get-buffer buffer))}}. In the test, the function @code{bufferp} returns true if its argument is a buffer---but false if its argument is the name of the buffer. (The last character of the function name @code{bufferp} is the character @samp{p}; as we saw earlier, such use of @samp{p} is a convention that indicates that the function is a predicate, which is a term that means that the function will determine whether some property is true or false. @xref{Wrong Type of Argument, , Using the Wrong Type Object as an Argument}.) @need 1200 The function @code{not} precedes the expression @code{(bufferp buffer)}, so the true-or-false-test looks like this: @smallexample (not (bufferp buffer)) @end smallexample @noindent @code{not} is a function that returns true if its argument is false and false if its argument is true. So if @code{(bufferp buffer)} returns true, the @code{not} expression returns false and vice versa. Using this test, the @code{if} expression works as follows: when the value of the variable @code{buffer} is actually a buffer rather than its name, the true-or-false-test returns false and the @code{if} expression does not evaluate the then-part. This is fine, since we do not need to do anything to the variable @code{buffer} if it really is a buffer. On the other hand, when the value of @code{buffer} is not a buffer itself, but the name of a buffer, the true-or-false-test returns true and the then-part of the expression is evaluated. In this case, the then-part is @code{(setq buffer (get-buffer buffer))}. This expression uses the @code{get-buffer} function to return an actual buffer itself, given its name. The @code{setq} then sets the variable @code{buffer} to the value of the buffer itself, replacing its previous value (which was the name of the buffer). @node Insert or @subsection The @code{or} in the Body The purpose of the @code{or} expression in the @code{insert-buffer} function is to ensure that the argument @code{buffer} is bound to a buffer and not just to the name of a buffer. The previous section shows how the job could have been done using an @code{if} expression. However, the @code{insert-buffer} function actually uses @code{or}. To understand this, it is necessary to understand how @code{or} works. @findex or An @code{or} function can have any number of arguments. It evaluates each argument in turn and returns the value of the first of its arguments that is not @code{nil}. Also, and this is a crucial feature of @code{or}, it does not evaluate any subsequent arguments after returning the first non-@code{nil} value. @need 800 The @code{or} expression looks like this: @smallexample @group (or (bufferp buffer) (setq buffer (get-buffer buffer))) @end group @end smallexample @noindent The first argument to @code{or} is the expression @code{(bufferp buffer)}. This expression returns true (a non-@code{nil} value) if the buffer is actually a buffer, and not just the name of a buffer. In the @code{or} expression, if this is the case, the @code{or} expression returns this true value and does not evaluate the next expression---and this is fine with us, since we do not want to do anything to the value of @code{buffer} if it really is a buffer. On the other hand, if the value of @code{(bufferp buffer)} is @code{nil}, which it will be if the value of @code{buffer} is the name of a buffer, the Lisp interpreter evaluates the next element of the @code{or} expression. This is the expression @code{(setq buffer (get-buffer buffer))}. This expression returns a non-@code{nil} value, which is the value to which it sets the variable @code{buffer}---and this value is a buffer itself, not the name of a buffer. The result of all this is that the symbol @code{buffer} is always bound to a buffer itself rather than to the name of a buffer. All this is necessary because the @code{set-buffer} function in a following line only works with a buffer itself, not with the name to a buffer. @need 1250 Incidentally, using @code{or}, the situation with the usher would be written like this: @smallexample (or (holding-on-to-guest) (find-and-take-arm-of-guest)) @end smallexample @node Insert let @subsection The @code{let} Expression in @code{insert-buffer} After ensuring that the variable @code{buffer} refers to a buffer itself and not just to the name of a buffer, the @code{insert-buffer function} continues with a @code{let} expression. This specifies three local variables, @code{start}, @code{end}, and @code{newmark} and binds them to the initial value @code{nil}. These variables are used inside the remainder of the @code{let} and temporarily hide any other occurrence of variables of the same name in Emacs until the end of the @code{let}. @need 1200 The body of the @code{let} contains two @code{save-excursion} expressions. First, we will look at the inner @code{save-excursion} expression in detail. The expression looks like this: @smallexample @group (save-excursion (set-buffer buffer) (setq start (point-min) end (point-max))) @end group @end smallexample @noindent The expression @code{(set-buffer buffer)} changes Emacs's attention from the current buffer to the one from which the text will copied. In that buffer, the variables @code{start} and @code{end} are set to the beginning and end of the buffer, using the commands @code{point-min} and @code{point-max}. Note that we have here an illustration of how @code{setq} is able to set two variables in the same expression. The first argument of @code{setq} is set to the value of its second, and its third argument is set to the value of its fourth. After the body of the inner @code{save-excursion} is evaluated, the @code{save-excursion} restores the original buffer, but @code{start} and @code{end} remain set to the values of the beginning and end of the buffer from which the text will be copied. @need 1250 The outer @code{save-excursion} expression looks like this: @smallexample @group (save-excursion (@var{inner-}@code{save-excursion}@var{-expression} (@var{go-to-new-buffer-and-set-}@code{start}@var{-and-}@code{end}) (insert-buffer-substring buffer start end) (setq newmark (point))) @end group @end smallexample @noindent The @code{insert-buffer-substring} function copies the text @emph{into} the current buffer @emph{from} the region indicated by @code{start} and @code{end} in @code{buffer}. Since the whole of the second buffer lies between @code{start} and @code{end}, the whole of the second buffer is copied into the buffer you are editing. Next, the value of point, which will be at the end of the inserted text, is recorded in the variable @code{newmark}. After the body of the outer @code{save-excursion} is evaluated, point is relocated to its original place. However, it is convenient to locate a mark at the end of the newly inserted text and locate point at its beginning. The @code{newmark} variable records the end of the inserted text. In the last line of the @code{let} expression, the @code{(push-mark newmark)} expression function sets a mark to this location. (The previous location of the mark is still accessible; it is recorded on the mark ring and you can go back to it with @kbd{C-u C-@key{SPC}}.) Meanwhile, point is located at the beginning of the inserted text, which is where it was before you called the insert function, the position of which was saved by the first @code{save-excursion}. @need 1250 The whole @code{let} expression looks like this: @smallexample @group (let (start end newmark) (save-excursion (save-excursion (set-buffer buffer) (setq start (point-min) end (point-max))) (insert-buffer-substring buffer start end) (setq newmark (point))) (push-mark newmark)) @end group @end smallexample Like the @code{append-to-buffer} function, the @code{insert-buffer} function uses @code{let}, @code{save-excursion}, and @code{set-buffer}. In addition, the function illustrates one way to use @code{or}. All these functions are building blocks that we will find and use again and again. @node New insert-buffer @subsection New Body for @code{insert-buffer} @findex insert-buffer@r{, new version body} @cindex new version body for @code{insert-buffer} The body in the GNU Emacs 22 version is more confusing than the original. @need 1250 It consists of two expressions, @smallexample @group (push-mark (save-excursion (insert-buffer-substring (get-buffer buffer)) (point))) nil @end group @end smallexample @noindent except, and this is what confuses novices, very important work is done inside the @code{push-mark} expression. The @code{get-buffer} function returns a buffer with the name provided. You will note that the function is @emph{not} called @code{get-buffer-create}; it does not create a buffer if one does not already exist. The buffer returned by @code{get-buffer}, an existing buffer, is passed to @code{insert-buffer-substring}, which inserts the whole of the buffer (since you did not specify anything else). The location into which the buffer is inserted is recorded by @code{push-mark}. Then the function returns @code{nil}, the value of its last command. Put another way, the @code{insert-buffer} function exists only to produce a side effect, inserting another buffer, not to return any value. @node beginning-of-buffer @section Complete Definition of @code{beginning-of-buffer} @findex beginning-of-buffer The basic structure of the @code{beginning-of-buffer} function has already been discussed. (@xref{simplified-beginning-of-buffer, , A Simplified @code{beginning-of-buffer} Definition}.) This section describes the complex part of the definition. As previously described, when invoked without an argument, @code{beginning-of-buffer} moves the cursor to the beginning of the buffer (in truth, the beginning of the accessible portion of the buffer), leaving the mark at the previous position. However, when the command is invoked with a number between one and ten, the function considers that number to be a fraction of the length of the buffer, measured in tenths, and Emacs moves the cursor that fraction of the way from the beginning of the buffer. Thus, you can either call this function with the key command @kbd{M-<}, which will move the cursor to the beginning of the buffer, or with a key command such as @kbd{C-u 7 M-<} which will move the cursor to a point 70% of the way through the buffer. If a number bigger than ten is used for the argument, it moves to the end of the buffer. The @code{beginning-of-buffer} function can be called with or without an argument. The use of the argument is optional. @menu * Optional Arguments:: * beginning-of-buffer opt arg:: Example with optional argument. * beginning-of-buffer complete:: @end menu @node Optional Arguments @subsection Optional Arguments Unless told otherwise, Lisp expects that a function with an argument in its function definition will be called with a value for that argument. If that does not happen, you get an error and a message that says @samp{Wrong number of arguments}. @cindex Optional arguments @cindex Keyword @findex optional However, optional arguments are a feature of Lisp: a particular @dfn{keyword} is used to tell the Lisp interpreter that an argument is optional. The keyword is @code{&optional}. (The @samp{&} in front of @samp{optional} is part of the keyword.) In a function definition, if an argument follows the keyword @code{&optional}, no value need be passed to that argument when the function is called. @need 1200 The first line of the function definition of @code{beginning-of-buffer} therefore looks like this: @smallexample (defun beginning-of-buffer (&optional arg) @end smallexample @need 1250 In outline, the whole function looks like this: @smallexample @group (defun beginning-of-buffer (&optional arg) "@var{documentation}@dots{}" (interactive "P") (or (@var{is-the-argument-a-cons-cell} arg) (and @var{are-both-transient-mark-mode-and-mark-active-true}) (push-mark)) (let (@var{determine-size-and-set-it}) (goto-char (@var{if-there-is-an-argument} @var{figure-out-where-to-go} @var{else-go-to} (point-min)))) @var{do-nicety} @end group @end smallexample The function is similar to the @code{simplified-beginning-of-buffer} function except that the @code{interactive} expression has @code{"P"} as an argument and the @code{goto-char} function is followed by an if-then-else expression that figures out where to put the cursor if there is an argument that is not a cons cell. (Since I do not explain a cons cell for many more chapters, please consider ignoring the function @code{consp}. @xref{List Implementation, , How Lists are Implemented}, and @ref{Cons Cell Type, , Cons Cell and List Types, elisp, The GNU Emacs Lisp Reference Manual}.) The @code{"P"} in the @code{interactive} expression tells Emacs to pass a prefix argument, if there is one, to the function in raw form. A prefix argument is made by typing the @key{META} key followed by a number, or by typing @kbd{C-u} and then a number. (If you don't type a number, @kbd{C-u} defaults to a cons cell with a 4. A lowercase @code{"p"} in the @code{interactive} expression causes the function to convert a prefix arg to a number.) The true-or-false-test of the @code{if} expression looks complex, but it is not: it checks whether @code{arg} has a value that is not @code{nil} and whether it is a cons cell. (That is what @code{consp} does; it checks whether its argument is a cons cell.) If @code{arg} has a value that is not @code{nil} (and is not a cons cell), which will be the case if @code{beginning-of-buffer} is called with a numeric argument, then this true-or-false-test will return true and the then-part of the @code{if} expression will be evaluated. On the other hand, if @code{beginning-of-buffer} is not called with an argument, the value of @code{arg} will be @code{nil} and the else-part of the @code{if} expression will be evaluated. The else-part is simply @code{point-min}, and when this is the outcome, the whole @code{goto-char} expression is @code{(goto-char (point-min))}, which is how we saw the @code{beginning-of-buffer} function in its simplified form. @node beginning-of-buffer opt arg @subsection @code{beginning-of-buffer} with an Argument When @code{beginning-of-buffer} is called with an argument, an expression is evaluated which calculates what value to pass to @code{goto-char}. This expression is rather complicated at first sight. It includes an inner @code{if} expression and much arithmetic. It looks like this: @smallexample @group (if (> (buffer-size) 10000) ;; @r{Avoid overflow for large buffer sizes!} (* (prefix-numeric-value arg) (/ size 10)) (/ (+ 10 (* size (prefix-numeric-value arg))) 10))) @end group @end smallexample @menu * Disentangle beginning-of-buffer:: * Large buffer case:: * Small buffer case:: @end menu @ifnottex @node Disentangle beginning-of-buffer @unnumberedsubsubsec Disentangle @code{beginning-of-buffer} @end ifnottex Like other complex-looking expressions, the conditional expression within @code{beginning-of-buffer} can be disentangled by looking at it as parts of a template, in this case, the template for an if-then-else expression. In skeletal form, the expression looks like this: @smallexample @group (if (@var{buffer-is-large} @var{divide-buffer-size-by-10-and-multiply-by-arg} @var{else-use-alternate-calculation} @end group @end smallexample The true-or-false-test of this inner @code{if} expression checks the size of the buffer. The reason for this is that the old version 18 Emacs used numbers that are no bigger than eight million or so and in the computation that followed, the programmer feared that Emacs might try to use over-large numbers if the buffer were large. The term ``overflow'', mentioned in the comment, means numbers that are over large. More recent versions of Emacs use larger numbers, but this code has not been touched, if only because people now look at buffers that are far, far larger than ever before. There are two cases: if the buffer is large and if it is not. @node Large buffer case @unnumberedsubsubsec What happens in a large buffer In @code{beginning-of-buffer}, the inner @code{if} expression tests whether the size of the buffer is greater than 10,000 characters. To do this, it uses the @code{>} function and the computation of @code{size} that comes from the let expression. In the old days, the function @code{buffer-size} was used. Not only was that function called several times, it gave the size of the whole buffer, not the accessible part. The computation makes much more sense when it handles just the accessible part. (@xref{Narrowing & Widening, , Narrowing and Widening}, for more information on focusing attention to an accessible part.) @need 800 The line looks like this: @smallexample (if (> size 10000) @end smallexample @need 1200 @noindent When the buffer is large, the then-part of the @code{if} expression is evaluated. It reads like this (after formatting for easy reading): @smallexample @group (* (prefix-numeric-value arg) (/ size 10)) @end group @end smallexample @noindent This expression is a multiplication, with two arguments to the function @code{*}. The first argument is @code{(prefix-numeric-value arg)}. When @code{"P"} is used as the argument for @code{interactive}, the value passed to the function as its argument is passed a @dfn{raw prefix argument}, and not a number. (It is a number in a list.) To perform the arithmetic, a conversion is necessary, and @code{prefix-numeric-value} does the job. @findex / @r{(division)} @cindex Division The second argument is @code{(/ size 10)}. This expression divides the numeric value by ten---the numeric value of the size of the accessible portion of the buffer. This produces a number that tells how many characters make up one tenth of the buffer size. (In Lisp, @code{/} is used for division, just as @code{*} is used for multiplication.) @need 1200 In the multiplication expression as a whole, this amount is multiplied by the value of the prefix argument---the multiplication looks like this: @smallexample @group (* @var{numeric-value-of-prefix-arg} @var{number-of-characters-in-one-tenth-of-the-accessible-buffer}) @end group @end smallexample @noindent If, for example, the prefix argument is @samp{7}, the one-tenth value will be multiplied by 7 to give a position 70% of the way through. @need 1200 The result of all this is that if the accessible portion of the buffer is large, the @code{goto-char} expression reads like this: @smallexample @group (goto-char (* (prefix-numeric-value arg) (/ size 10))) @end group @end smallexample This puts the cursor where we want it. @node Small buffer case @unnumberedsubsubsec What happens in a small buffer If the buffer contains fewer than 10,000 characters, a slightly different computation is performed. You might think this is not necessary, since the first computation could do the job. However, in a small buffer, the first method may not put the cursor on exactly the desired line; the second method does a better job. @need 800 The code looks like this: @c Keep this on one line. @smallexample (/ (+ 10 (* size (prefix-numeric-value arg))) 10)) @end smallexample @need 1200 @noindent This is code in which you figure out what happens by discovering how the functions are embedded in parentheses. It is easier to read if you reformat it with each expression indented more deeply than its enclosing expression: @smallexample @group (/ (+ 10 (* size (prefix-numeric-value arg))) 10)) @end group @end smallexample @need 1200 @noindent Looking at parentheses, we see that the innermost operation is @code{(prefix-numeric-value arg)}, which converts the raw argument to a number. In the following expression, this number is multiplied by the size of the accessible portion of the buffer: @smallexample (* size (prefix-numeric-value arg)) @end smallexample @noindent This multiplication creates a number that may be larger than the size of the buffer---seven times larger if the argument is 7, for example. Ten is then added to this number and finally the large number is divided by ten to provide a value that is one character larger than the percentage position in the buffer. The number that results from all this is passed to @code{goto-char} and the cursor is moved to that point. @need 1500 @node beginning-of-buffer complete @subsection The Complete @code{beginning-of-buffer} @need 1000 Here is the complete text of the @code{beginning-of-buffer} function: @sp 1 @c In GNU Emacs 22 @smallexample @group (defun beginning-of-buffer (&optional arg) "Move point to the beginning of the buffer; leave mark at previous position. With \\[universal-argument] prefix, do not set mark at previous position. With numeric arg N, put point N/10 of the way from the beginning. If the buffer is narrowed, this command uses the beginning and size of the accessible part of the buffer. @end group @group Don't use this command in Lisp programs! \(goto-char (point-min)) is faster and avoids clobbering the mark." (interactive "P") (or (consp arg) (and transient-mark-mode mark-active) (push-mark)) @end group @group (let ((size (- (point-max) (point-min)))) (goto-char (if (and arg (not (consp arg))) (+ (point-min) (if (> size 10000) ;; Avoid overflow for large buffer sizes! (* (prefix-numeric-value arg) (/ size 10)) (/ (+ 10 (* size (prefix-numeric-value arg))) 10))) (point-min)))) (if (and arg (not (consp arg))) (forward-line 1))) @end group @end smallexample @ignore From before GNU Emacs 22 @smallexample @group (defun beginning-of-buffer (&optional arg) "Move point to the beginning of the buffer; leave mark at previous position. With arg N, put point N/10 of the way from the true beginning. @end group @group Don't use this in Lisp programs! \(goto-char (point-min)) is faster and does not set the mark." (interactive "P") (push-mark) @end group @group (goto-char (if arg (if (> (buffer-size) 10000) ;; @r{Avoid overflow for large buffer sizes!} (* (prefix-numeric-value arg) (/ (buffer-size) 10)) @end group @group (/ (+ 10 (* (buffer-size) (prefix-numeric-value arg))) 10)) (point-min))) (if arg (forward-line 1))) @end group @end smallexample @end ignore @noindent Except for two small points, the previous discussion shows how this function works. The first point deals with a detail in the documentation string, and the second point concerns the last line of the function. @need 800 In the documentation string, there is reference to an expression: @smallexample \\[universal-argument] @end smallexample @noindent A @samp{\\} is used before the first square bracket of this expression. This @samp{\\} tells the Lisp interpreter to substitute whatever key is currently bound to the @samp{[@dots{}]}. In the case of @code{universal-argument}, that is usually @kbd{C-u}, but it might be different. (@xref{Documentation Tips, , Tips for Documentation Strings, elisp, The GNU Emacs Lisp Reference Manual}, for more information.) @need 1200 Finally, the last line of the @code{beginning-of-buffer} command says to move point to the beginning of the next line if the command is invoked with an argument: @smallexample (if (and arg (not (consp arg))) (forward-line 1)) @end smallexample @noindent This puts the cursor at the beginning of the first line after the appropriate tenths position in the buffer. This is a flourish that means that the cursor is always located @emph{at least} the requested tenths of the way through the buffer, which is a nicety that is, perhaps, not necessary, but which, if it did not occur, would be sure to draw complaints. (The @code{(not (consp arg))} portion is so that if you specify the command with a @kbd{C-u}, but without a number, that is to say, if the raw prefix argument is simply a cons cell, the command does not put you at the beginning of the second line.) @node Second Buffer Related Review @section Review Here is a brief summary of some of the topics covered in this chapter. @table @code @item or Evaluate each argument in sequence, and return the value of the first argument that is not @code{nil}; if none return a value that is not @code{nil}, return @code{nil}. In brief, return the first true value of the arguments; return a true value if one @emph{or} any of the others are true. @item and Evaluate each argument in sequence, and if any are @code{nil}, return @code{nil}; if none are @code{nil}, return the value of the last argument. In brief, return a true value only if all the arguments are true; return a true value if one @emph{and} each of the others is true. @item &optional A keyword used to indicate that an argument to a function definition is optional; this means that the function can be evaluated without the argument, if desired. @item prefix-numeric-value Convert the raw prefix argument produced by @code{(interactive "P")} to a numeric value. @item forward-line Move point forward to the beginning of the next line, or if the argument is greater than one, forward that many lines. If it can't move as far forward as it is supposed to, @code{forward-line} goes forward as far as it can and then returns a count of the number of additional lines it was supposed to move but couldn't. @item erase-buffer Delete the entire contents of the current buffer. @item bufferp Return @code{t} if its argument is a buffer; otherwise return @code{nil}. @end table @node optional Exercise @section @code{optional} Argument Exercise Write an interactive function with an optional argument that tests whether its argument, a number, is greater than or equal to, or else, less than the value of @code{fill-column}, and tells you which, in a message. However, if you do not pass an argument to the function, use 56 as a default value. @node Narrowing & Widening @chapter Narrowing and Widening @cindex Focusing attention (narrowing) @cindex Narrowing @cindex Widening Narrowing is a feature of Emacs that makes it possible for you to focus on a specific part of a buffer, and work without accidentally changing other parts. Narrowing is normally disabled since it can confuse novices. @menu * Narrowing advantages:: The advantages of narrowing * save-restriction:: The @code{save-restriction} special form. * what-line:: The number of the line that point is on. * narrow Exercise:: @end menu @ifnottex @node Narrowing advantages @unnumberedsec The Advantages of Narrowing @end ifnottex With narrowing, the rest of a buffer is made invisible, as if it weren't there. This is an advantage if, for example, you want to replace a word in one part of a buffer but not in another: you narrow to the part you want and the replacement is carried out only in that section, not in the rest of the buffer. Searches will only work within a narrowed region, not outside of one, so if you are fixing a part of a document, you can keep yourself from accidentally finding parts you do not need to fix by narrowing just to the region you want. (The key binding for @code{narrow-to-region} is @kbd{C-x n n}.) However, narrowing does make the rest of the buffer invisible, which can scare people who inadvertently invoke narrowing and think they have deleted a part of their file. Moreover, the @code{undo} command (which is usually bound to @kbd{C-x u}) does not turn off narrowing (nor should it), so people can become quite desperate if they do not know that they can return the rest of a buffer to visibility with the @code{widen} command. (The key binding for @code{widen} is @kbd{C-x n w}.) Narrowing is just as useful to the Lisp interpreter as to a human. Often, an Emacs Lisp function is designed to work on just part of a buffer; or conversely, an Emacs Lisp function needs to work on all of a buffer that has been narrowed. The @code{what-line} function, for example, removes the narrowing from a buffer, if it has any narrowing and when it has finished its job, restores the narrowing to what it was. On the other hand, the @code{count-lines} function uses narrowing to restrict itself to just that portion of the buffer in which it is interested and then restores the previous situation. @node save-restriction @section The @code{save-restriction} Special Form @findex save-restriction In Emacs Lisp, you can use the @code{save-restriction} special form to keep track of whatever narrowing is in effect, if any. When the Lisp interpreter meets with @code{save-restriction}, it executes the code in the body of the @code{save-restriction} expression, and then undoes any changes to narrowing that the code caused. If, for example, the buffer is narrowed and the code that follows @code{save-restriction} gets rid of the narrowing, @code{save-restriction} returns the buffer to its narrowed region afterwards. In the @code{what-line} command, any narrowing the buffer may have is undone by the @code{widen} command that immediately follows the @code{save-restriction} command. Any original narrowing is restored just before the completion of the function. @need 1250 The template for a @code{save-restriction} expression is simple: @smallexample @group (save-restriction @var{body}@dots{} ) @end group @end smallexample @noindent The body of the @code{save-restriction} is one or more expressions that will be evaluated in sequence by the Lisp interpreter. Finally, a point to note: when you use both @code{save-excursion} and @code{save-restriction}, one right after the other, you should use @code{save-excursion} outermost. If you write them in reverse order, you may fail to record narrowing in the buffer to which Emacs switches after calling @code{save-excursion}. Thus, when written together, @code{save-excursion} and @code{save-restriction} should be written like this: @smallexample @group (save-excursion (save-restriction @var{body}@dots{})) @end group @end smallexample In other circumstances, when not written together, the @code{save-excursion} and @code{save-restriction} special forms must be written in the order appropriate to the function. @need 1250 For example, @smallexample @group (save-restriction (widen) (save-excursion @var{body}@dots{})) @end group @end smallexample @ignore Emacs 22 /usr/local/src/emacs/lisp/simple.el (defun what-line () "Print the current buffer line number and narrowed line number of point." (interactive) (let ((start (point-min)) (n (line-number-at-pos))) (if (= start 1) (message "Line %d" n) (save-excursion (save-restriction (widen) (message "line %d (narrowed line %d)" (+ n (line-number-at-pos start) -1) n)))))) (defun line-number-at-pos (&optional pos) "Return (narrowed) buffer line number at position POS. If POS is nil, use current buffer location. Counting starts at (point-min), so the value refers to the contents of the accessible portion of the buffer." (let ((opoint (or pos (point))) start) (save-excursion (goto-char (point-min)) (setq start (point)) (goto-char opoint) (forward-line 0) (1+ (count-lines start (point)))))) (defun count-lines (start end) "Return number of lines between START and END. This is usually the number of newlines between them, but can be one more if START is not equal to END and the greater of them is not at the start of a line." (save-excursion (save-restriction (narrow-to-region start end) (goto-char (point-min)) (if (eq selective-display t) (save-match-data (let ((done 0)) (while (re-search-forward "[\n\C-m]" nil t 40) (setq done (+ 40 done))) (while (re-search-forward "[\n\C-m]" nil t 1) (setq done (+ 1 done))) (goto-char (point-max)) (if (and (/= start end) (not (bolp))) (1+ done) done))) (- (buffer-size) (forward-line (buffer-size))))))) @end ignore @node what-line @section @code{what-line} @findex what-line @cindex Widening, example of The @code{what-line} command tells you the number of the line in which the cursor is located. The function illustrates the use of the @code{save-restriction} and @code{save-excursion} commands. Here is the original text of the function: @smallexample @group (defun what-line () "Print the current line number (in the buffer) of point." (interactive) (save-restriction (widen) (save-excursion (beginning-of-line) (message "Line %d" (1+ (count-lines 1 (point))))))) @end group @end smallexample (In recent versions of GNU Emacs, the @code{what-line} function has been expanded to tell you your line number in a narrowed buffer as well as your line number in a widened buffer. The recent version is more complex than the version shown here. If you feel adventurous, you might want to look at it after figuring out how this version works. You will probably need to use @kbd{C-h f} (@code{describe-function}). The newer version uses a conditional to determine whether the buffer has been narrowed. (Also, it uses @code{line-number-at-pos}, which among other simple expressions, such as @code{(goto-char (point-min))}, moves point to the beginning of the current line with @code{(forward-line 0)} rather than @code{beginning-of-line}.) The @code{what-line} function as shown here has a documentation line and is interactive, as you would expect. The next two lines use the functions @code{save-restriction} and @code{widen}. The @code{save-restriction} special form notes whatever narrowing is in effect, if any, in the current buffer and restores that narrowing after the code in the body of the @code{save-restriction} has been evaluated. The @code{save-restriction} special form is followed by @code{widen}. This function undoes any narrowing the current buffer may have had when @code{what-line} was called. (The narrowing that was there is the narrowing that @code{save-restriction} remembers.) This widening makes it possible for the line counting commands to count from the beginning of the buffer. Otherwise, they would have been limited to counting within the accessible region. Any original narrowing is restored just before the completion of the function by the @code{save-restriction} special form. The call to @code{widen} is followed by @code{save-excursion}, which saves the location of the cursor (i.e., of point), and restores it after the code in the body of the @code{save-excursion} uses the @code{beginning-of-line} function to move point. (Note that the @code{(widen)} expression comes between the @code{save-restriction} and @code{save-excursion} special forms. When you write the two @code{save- @dots{}} expressions in sequence, write @code{save-excursion} outermost.) @need 1200 The last two lines of the @code{what-line} function are functions to count the number of lines in the buffer and then print the number in the echo area. @smallexample @group (message "Line %d" (1+ (count-lines 1 (point))))))) @end group @end smallexample The @code{message} function prints a one-line message at the bottom of the Emacs screen. The first argument is inside of quotation marks and is printed as a string of characters. However, it may contain a @samp{%d} expression to print a following argument. @samp{%d} prints the argument as a decimal, so the message will say something such as @samp{Line 243}. @need 1200 The number that is printed in place of the @samp{%d} is computed by the last line of the function: @smallexample (1+ (count-lines 1 (point))) @end smallexample @ignore GNU Emacs 22 (defun count-lines (start end) "Return number of lines between START and END. This is usually the number of newlines between them, but can be one more if START is not equal to END and the greater of them is not at the start of a line." (save-excursion (save-restriction (narrow-to-region start end) (goto-char (point-min)) (if (eq selective-display t) (save-match-data (let ((done 0)) (while (re-search-forward "[\n\C-m]" nil t 40) (setq done (+ 40 done))) (while (re-search-forward "[\n\C-m]" nil t 1) (setq done (+ 1 done))) (goto-char (point-max)) (if (and (/= start end) (not (bolp))) (1+ done) done))) (- (buffer-size) (forward-line (buffer-size))))))) @end ignore @noindent What this does is count the lines from the first position of the buffer, indicated by the @code{1}, up to @code{(point)}, and then add one to that number. (The @code{1+} function adds one to its argument.) We add one to it because line 2 has only one line before it, and @code{count-lines} counts only the lines @emph{before} the current line. After @code{count-lines} has done its job, and the message has been printed in the echo area, the @code{save-excursion} restores point to its original position; and @code{save-restriction} restores the original narrowing, if any. @node narrow Exercise @section Exercise with Narrowing Write a function that will display the first 60 characters of the current buffer, even if you have narrowed the buffer to its latter half so that the first line is inaccessible. Restore point, mark, and narrowing. For this exercise, you need to use a whole potpourri of functions, including @code{save-restriction}, @code{widen}, @code{goto-char}, @code{point-min}, @code{message}, and @code{buffer-substring}. @cindex Properties, mention of @code{buffer-substring-no-properties} (@code{buffer-substring} is a previously unmentioned function you will have to investigate yourself; or perhaps you will have to use @code{buffer-substring-no-properties} or @code{filter-buffer-substring} @dots{}, yet other functions. Text properties are a feature otherwise not discussed here. @xref{Text Properties, , Text Properties, elisp, The GNU Emacs Lisp Reference Manual}.) Additionally, do you really need @code{goto-char} or @code{point-min}? Or can you write the function without them? @node car cdr & cons @chapter @code{car}, @code{cdr}, @code{cons}: Fundamental Functions @findex car@r{, introduced} @findex cdr@r{, introduced} In Lisp, @code{car}, @code{cdr}, and @code{cons} are fundamental functions. The @code{cons} function is used to construct lists, and the @code{car} and @code{cdr} functions are used to take them apart. In the walk through of the @code{copy-region-as-kill} function, we will see @code{cons} as well as two variants on @code{cdr}, namely, @code{setcdr} and @code{nthcdr}. (@xref{copy-region-as-kill}.) @menu * Strange Names:: A historical aside: why the strange names? * car & cdr:: Functions for extracting part of a list. * cons:: Constructing a list. * nthcdr:: Calling @code{cdr} repeatedly. * nth:: * setcar:: Changing the first element of a list. * setcdr:: Changing the rest of a list. * cons Exercise:: @end menu @ifnottex @node Strange Names @unnumberedsec Strange Names @end ifnottex The name of the @code{cons} function is not unreasonable: it is an abbreviation of the word ``construct''. The origins of the names for @code{car} and @code{cdr}, on the other hand, are esoteric: @code{car} is an acronym from the phrase ``Contents of the Address part of the Register''; and @code{cdr} (pronounced ``could-er'') is an acronym from the phrase ``Contents of the Decrement part of the Register''. These phrases refer to specific pieces of hardware on the very early computer on which the original Lisp was developed. Besides being obsolete, the phrases have been completely irrelevant for more than 25 years to anyone thinking about Lisp. Nonetheless, although a few brave scholars have begun to use more reasonable names for these functions, the old terms are still in use. In particular, since the terms are used in the Emacs Lisp source code, we will use them in this introduction. @node car & cdr @section @code{car} and @code{cdr} The @sc{car} of a list is, quite simply, the first item in the list. Thus the @sc{car} of the list @code{(rose violet daisy buttercup)} is @code{rose}. @need 1200 If you are reading this in Info in GNU Emacs, you can see this by evaluating the following: @smallexample (car '(rose violet daisy buttercup)) @end smallexample @noindent After evaluating the expression, @code{rose} will appear in the echo area. Clearly, a more reasonable name for the @code{car} function would be @code{first} and this is often suggested. @code{car} does not remove the first item from the list; it only reports what it is. After @code{car} has been applied to a list, the list is still the same as it was. In the jargon, @code{car} is ``non-destructive''. This feature turns out to be important. The @sc{cdr} of a list is the rest of the list, that is, the @code{cdr} function returns the part of the list that follows the first item. Thus, while the @sc{car} of the list @code{'(rose violet daisy buttercup)} is @code{rose}, the rest of the list, the value returned by the @code{cdr} function, is @code{(violet daisy buttercup)}. @need 800 You can see this by evaluating the following in the usual way: @smallexample (cdr '(rose violet daisy buttercup)) @end smallexample @noindent When you evaluate this, @code{(violet daisy buttercup)} will appear in the echo area. Like @code{car}, @code{cdr} does not remove any elements from the list---it just returns a report of what the second and subsequent elements are. Incidentally, in the example, the list of flowers is quoted. If it were not, the Lisp interpreter would try to evaluate the list by calling @code{rose} as a function. In this example, we do not want to do that. Clearly, a more reasonable name for @code{cdr} would be @code{rest}. (There is a lesson here: when you name new functions, consider very carefully what you are doing, since you may be stuck with the names for far longer than you expect. The reason this document perpetuates these names is that the Emacs Lisp source code uses them, and if I did not use them, you would have a hard time reading the code; but do, please, try to avoid using these terms yourself. The people who come after you will be grateful to you.) When @code{car} and @code{cdr} are applied to a list made up of symbols, such as the list @code{(pine fir oak maple)}, the element of the list returned by the function @code{car} is the symbol @code{pine} without any parentheses around it. @code{pine} is the first element in the list. However, the @sc{cdr} of the list is a list itself, @code{(fir oak maple)}, as you can see by evaluating the following expressions in the usual way: @smallexample @group (car '(pine fir oak maple)) (cdr '(pine fir oak maple)) @end group @end smallexample On the other hand, in a list of lists, the first element is itself a list. @code{car} returns this first element as a list. For example, the following list contains three sub-lists, a list of carnivores, a list of herbivores and a list of sea mammals: @smallexample @group (car '((lion tiger cheetah) (gazelle antelope zebra) (whale dolphin seal))) @end group @end smallexample @noindent In this example, the first element or @sc{car} of the list is the list of carnivores, @code{(lion tiger cheetah)}, and the rest of the list is @code{((gazelle antelope zebra) (whale dolphin seal))}. @smallexample @group (cdr '((lion tiger cheetah) (gazelle antelope zebra) (whale dolphin seal))) @end group @end smallexample It is worth saying again that @code{car} and @code{cdr} are non-destructive---that is, they do not modify or change lists to which they are applied. This is very important for how they are used. Also, in the first chapter, in the discussion about atoms, I said that in Lisp, certain kinds of atom, such as an array, can be separated into parts; but the mechanism for doing this is different from the mechanism for splitting a list. As far as Lisp is concerned, the atoms of a list are unsplittable. (@xref{Lisp Atoms}.) The @code{car} and @code{cdr} functions are used for splitting lists and are considered fundamental to Lisp. Since they cannot split or gain access to the parts of an array, an array is considered an atom. Conversely, the other fundamental function, @code{cons}, can put together or construct a list, but not an array. (Arrays are handled by array-specific functions. @xref{Arrays, , Arrays, elisp, The GNU Emacs Lisp Reference Manual}.) @node cons @section @code{cons} @findex cons@r{, introduced} The @code{cons} function constructs lists; it is the inverse of @code{car} and @code{cdr}. For example, @code{cons} can be used to make a four element list from the three element list, @code{(fir oak maple)}: @smallexample (cons 'pine '(fir oak maple)) @end smallexample @need 800 @noindent After evaluating this list, you will see @smallexample (pine fir oak maple) @end smallexample @noindent appear in the echo area. @code{cons} causes the creation of a new list in which the element is followed by the elements of the original list. We often say that @code{cons} puts a new element at the beginning of a list, or that it attaches or pushes elements onto the list, but this phrasing can be misleading, since @code{cons} does not change an existing list, but creates a new one. Like @code{car} and @code{cdr}, @code{cons} is non-destructive. @menu * Build a list:: * length:: How to find the length of a list. @end menu @ifnottex @node Build a list @unnumberedsubsec Build a list @end ifnottex @code{cons} must have a list to attach to.@footnote{Actually, you can @code{cons} an element to an atom to produce a dotted pair. Dotted pairs are not discussed here; see @ref{Dotted Pair Notation, , Dotted Pair Notation, elisp, The GNU Emacs Lisp Reference Manual}.} You cannot start from absolutely nothing. If you are building a list, you need to provide at least an empty list at the beginning. Here is a series of @code{cons} expressions that build up a list of flowers. If you are reading this in Info in GNU Emacs, you can evaluate each of the expressions in the usual way; the value is printed in this text after @samp{@result{}}, which you may read as ``evaluates to''. @smallexample @group (cons 'buttercup ()) @result{} (buttercup) @end group @group (cons 'daisy '(buttercup)) @result{} (daisy buttercup) @end group @group (cons 'violet '(daisy buttercup)) @result{} (violet daisy buttercup) @end group @group (cons 'rose '(violet daisy buttercup)) @result{} (rose violet daisy buttercup) @end group @end smallexample @noindent In the first example, the empty list is shown as @code{()} and a list made up of @code{buttercup} followed by the empty list is constructed. As you can see, the empty list is not shown in the list that was constructed. All that you see is @code{(buttercup)}. The empty list is not counted as an element of a list because there is nothing in an empty list. Generally speaking, an empty list is invisible. The second example, @code{(cons 'daisy '(buttercup))} constructs a new, two element list by putting @code{daisy} in front of @code{buttercup}; and the third example constructs a three element list by putting @code{violet} in front of @code{daisy} and @code{buttercup}. @node length @subsection Find the Length of a List: @code{length} @findex length You can find out how many elements there are in a list by using the Lisp function @code{length}, as in the following examples: @smallexample @group (length '(buttercup)) @result{} 1 @end group @group (length '(daisy buttercup)) @result{} 2 @end group @group (length (cons 'violet '(daisy buttercup))) @result{} 3 @end group @end smallexample @noindent In the third example, the @code{cons} function is used to construct a three element list which is then passed to the @code{length} function as its argument. @need 1200 We can also use @code{length} to count the number of elements in an empty list: @smallexample @group (length ()) @result{} 0 @end group @end smallexample @noindent As you would expect, the number of elements in an empty list is zero. An interesting experiment is to find out what happens if you try to find the length of no list at all; that is, if you try to call @code{length} without giving it an argument, not even an empty list: @smallexample (length ) @end smallexample @need 800 @noindent What you see, if you evaluate this, is the error message @smallexample Lisp error: (wrong-number-of-arguments length 0) @end smallexample @noindent This means that the function receives the wrong number of arguments, zero, when it expects some other number of arguments. In this case, one argument is expected, the argument being a list whose length the function is measuring. (Note that @emph{one} list is @emph{one} argument, even if the list has many elements inside it.) The part of the error message that says @samp{length} is the name of the function. @ignore @code{length} is still a subroutine, but you need C-h f to discover that. In an earlier version: This is written with a special notation, @samp{# (length string) 0) (if yank-handler (put-text-property 0 (length string) 'yank-handler yank-handler string)) (if yank-handler (signal 'args-out-of-range (list string "yank-handler specified for empty string")))) @end group @group (if (fboundp 'menu-bar-update-yank-menu) (menu-bar-update-yank-menu string (and replace (car kill-ring)))) @end group @group (if (and replace kill-ring) (setcar kill-ring string) (push string kill-ring) (if (> (length kill-ring) kill-ring-max) (setcdr (nthcdr (1- kill-ring-max) kill-ring) nil))) @end group @group (setq kill-ring-yank-pointer kill-ring) (if interprogram-cut-function (funcall interprogram-cut-function string (not replace)))) @end group @end smallexample @ignore was: (defun kill-new (string &optional replace) "Make STRING the latest kill in the kill ring. Set the kill-ring-yank pointer to point to it. If `interprogram-cut-function' is non-nil, apply it to STRING. Optional second argument REPLACE non-nil means that STRING will replace the front of the kill ring, rather than being added to the list." (and (fboundp 'menu-bar-update-yank-menu) (menu-bar-update-yank-menu string (and replace (car kill-ring)))) (if (and replace kill-ring) (setcar kill-ring string) (setq kill-ring (cons string kill-ring)) (if (> (length kill-ring) kill-ring-max) (setcdr (nthcdr (1- kill-ring-max) kill-ring) nil))) (setq kill-ring-yank-pointer kill-ring) (if interprogram-cut-function (funcall interprogram-cut-function string (not replace)))) @end ignore (Notice that the function is not interactive.) As usual, we can look at this function in parts. The function definition has an optional @code{yank-handler} argument, which when invoked tells the function how to deal with properties added to the text, such as bold or italics. We will skip that. @need 1200 The first line of the documentation makes sense: @smallexample Make STRING the latest kill in the kill ring. @end smallexample @noindent Let's skip over the rest of the documentation for the moment. @noindent Also, let's skip over the initial @code{if} expression and those lines of code involving @code{menu-bar-update-yank-menu}. We will explain them below. @need 1200 The critical lines are these: @smallexample @group (if (and replace kill-ring) ;; @r{then} (setcar kill-ring string) @end group @group ;; @r{else} (push string kill-ring) @end group @group (if (> (length kill-ring) kill-ring-max) ;; @r{avoid overly long kill ring} (setcdr (nthcdr (1- kill-ring-max) kill-ring) nil))) @end group @group (setq kill-ring-yank-pointer kill-ring) (if interprogram-cut-function (funcall interprogram-cut-function string (not replace)))) @end group @end smallexample The conditional test is @w{@code{(and replace kill-ring)}}. This will be true when two conditions are met: the kill ring has something in it, and the @code{replace} variable is true. @need 1250 When the @code{kill-append} function sets @code{replace} to be true and when the kill ring has at least one item in it, the @code{setcar} expression is executed: @smallexample (setcar kill-ring string) @end smallexample The @code{setcar} function actually changes the first element of the @code{kill-ring} list to the value of @code{string}. It replaces the first element. @need 1250 On the other hand, if the kill ring is empty, or replace is false, the else-part of the condition is executed: @smallexample (push string kill-ring) @end smallexample @noindent @need 1250 @code{push} puts its first argument onto the second. It is similar to the older @smallexample (setq kill-ring (cons string kill-ring)) @end smallexample @noindent @need 1250 or the newer @smallexample (add-to-list kill-ring string) @end smallexample @noindent When it is false, the expression first constructs a new version of the kill ring by prepending @code{string} to the existing kill ring as a new element (that is what the @code{push} does). Then it executes a second @code{if} clause. This second @code{if} clause keeps the kill ring from growing too long. Let's look at these two expressions in order. The @code{push} line of the else-part sets the new value of the kill ring to what results from adding the string being killed to the old kill ring. We can see how this works with an example. @need 800 First, @smallexample (setq example-list '("here is a clause" "another clause")) @end smallexample @need 1200 @noindent After evaluating this expression with @kbd{C-x C-e}, you can evaluate @code{example-list} and see what it returns: @smallexample @group example-list @result{} ("here is a clause" "another clause") @end group @end smallexample @need 1200 @noindent Now, we can add a new element on to this list by evaluating the following expression: @findex push@r{, example} @smallexample (push "a third clause" example-list) @end smallexample @need 800 @noindent When we evaluate @code{example-list}, we find its value is: @smallexample @group example-list @result{} ("a third clause" "here is a clause" "another clause") @end group @end smallexample @noindent Thus, the third clause is added to the list by @code{push}. @need 1200 Now for the second part of the @code{if} clause. This expression keeps the kill ring from growing too long. It looks like this: @smallexample @group (if (> (length kill-ring) kill-ring-max) (setcdr (nthcdr (1- kill-ring-max) kill-ring) nil)) @end group @end smallexample The code checks whether the length of the kill ring is greater than the maximum permitted length. This is the value of @code{kill-ring-max} (which is 60, by default). If the length of the kill ring is too long, then this code sets the last element of the kill ring to @code{nil}. It does this by using two functions, @code{nthcdr} and @code{setcdr}. We looked at @code{setcdr} earlier (@pxref{setcdr, , @code{setcdr}}). It sets the @sc{cdr} of a list, just as @code{setcar} sets the @sc{car} of a list. In this case, however, @code{setcdr} will not be setting the @sc{cdr} of the whole kill ring; the @code{nthcdr} function is used to cause it to set the @sc{cdr} of the next to last element of the kill ring---this means that since the @sc{cdr} of the next to last element is the last element of the kill ring, it will set the last element of the kill ring. @findex nthcdr@r{, example} The @code{nthcdr} function works by repeatedly taking the @sc{cdr} of a list---it takes the @sc{cdr} of the @sc{cdr} of the @sc{cdr} @dots{} It does this @var{N} times and returns the results. (@xref{nthcdr, , @code{nthcdr}}.) @findex setcdr@r{, example} Thus, if we had a four element list that was supposed to be three elements long, we could set the @sc{cdr} of the next to last element to @code{nil}, and thereby shorten the list. (If you set the last element to some other value than @code{nil}, which you could do, then you would not have shortened the list. @xref{setcdr, , @code{setcdr}}.) You can see shortening by evaluating the following three expressions in turn. First set the value of @code{trees} to @code{(maple oak pine birch)}, then set the @sc{cdr} of its second @sc{cdr} to @code{nil} and then find the value of @code{trees}: @smallexample @group (setq trees (list 'maple 'oak 'pine 'birch)) @result{} (maple oak pine birch) @end group @group (setcdr (nthcdr 2 trees) nil) @result{} nil trees @result{} (maple oak pine) @end group @end smallexample @noindent (The value returned by the @code{setcdr} expression is @code{nil} since that is what the @sc{cdr} is set to.) To repeat, in @code{kill-new}, the @code{nthcdr} function takes the @sc{cdr} a number of times that is one less than the maximum permitted size of the kill ring and @code{setcdr} sets the @sc{cdr} of that element (which will be the rest of the elements in the kill ring) to @code{nil}. This prevents the kill ring from growing too long. @need 800 The next to last expression in the @code{kill-new} function is @smallexample (setq kill-ring-yank-pointer kill-ring) @end smallexample The @code{kill-ring-yank-pointer} is a global variable that is set to be the @code{kill-ring}. Even though the @code{kill-ring-yank-pointer} is called a @samp{pointer}, it is a variable just like the kill ring. However, the name has been chosen to help humans understand how the variable is used. @need 1200 Now, to return to an early expression in the body of the function: @smallexample @group (if (fboundp 'menu-bar-update-yank-menu) (menu-bar-update-yank-menu string (and replace (car kill-ring)))) @end group @end smallexample @noindent It starts with an @code{if} expression In this case, the expression tests first to see whether @code{menu-bar-update-yank-menu} exists as a function, and if so, calls it. The @code{fboundp} function returns true if the symbol it is testing has a function definition that is not void. If the symbol's function definition were void, we would receive an error message, as we did when we created errors intentionally (@pxref{Making Errors, , Generate an Error Message}). @noindent The then-part contains an expression whose first element is the function @code{and}. @findex and The @code{and} special form evaluates each of its arguments until one of the arguments returns a value of @code{nil}, in which case the @code{and} expression returns @code{nil}; however, if none of the arguments returns a value of @code{nil}, the value resulting from evaluating the last argument is returned. (Since such a value is not @code{nil}, it is considered true in Emacs Lisp.) In other words, an @code{and} expression returns a true value only if all its arguments are true. (@xref{Second Buffer Related Review}.) The expression determines whether the second argument to @code{menu-bar-update-yank-menu} is true or not. @ignore ;; If we're supposed to be extending an existing string, and that ;; string really is at the front of the menu, then update it in place. @end ignore @code{menu-bar-update-yank-menu} is one of the functions that make it possible to use the ``Select and Paste'' menu in the Edit item of a menu bar; using a mouse, you can look at the various pieces of text you have saved and select one piece to paste. The last expression in the @code{kill-new} function adds the newly copied string to whatever facility exists for copying and pasting among different programs running in a windowing system. In the X Windowing system, for example, the @code{x-select-text} function takes the string and stores it in memory operated by X@. You can paste the string in another program, such as an Xterm. @need 1200 The expression looks like this: @smallexample @group (if interprogram-cut-function (funcall interprogram-cut-function string (not replace)))) @end group @end smallexample If an @code{interprogram-cut-function} exists, then Emacs executes @code{funcall}, which in turn calls its first argument as a function and passes the remaining arguments to it. (Incidentally, as far as I can see, this @code{if} expression could be replaced by an @code{and} expression similar to the one in the first part of the function.) We are not going to discuss windowing systems and other programs further, but merely note that this is a mechanism that enables GNU Emacs to work easily and well with other programs. This code for placing text in the kill ring, either concatenated with an existing element or as a new element, leads us to the code for bringing back text that has been cut out of the buffer---the yank commands. However, before discussing the yank commands, it is better to learn how lists are implemented in a computer. This will make clear such mysteries as the use of the term ``pointer''. But before that, we will digress into C. @ignore @c is this true in Emacs 22? Does not seems to be (If the @w{@code{(< end beg))}} expression is true, @code{kill-append} prepends the string to the just previously clipped text. For a detailed discussion, see @ref{kill-append function, , The @code{kill-append} function}.) If you then yank back the text, i.e., paste it, you get both pieces of text at once. That way, if you delete two words in a row, and then yank them back, you get both words, in their proper order, with one yank. (The @w{@code{(< end beg))}} expression makes sure the order is correct.) On the other hand, if the previous command is not @code{kill-region}, then the @code{kill-new} function is called, which adds the text to the kill ring as the latest item, and sets the @code{kill-ring-yank-pointer} variable to point to it. @end ignore @ignore @c Evidently, changed for Emacs 22. The zap-to-char command does not @c use the delete-and-extract-region function 2006 Oct 26, the Digression into C is now OK but should come after copy-region-as-kill and filter-buffer-substring 2006 Oct 24 In Emacs 22, copy-region-as-kill is short, 12 lines, and uses filter-buffer-substring, which is longer, 39 lines and has delete-and-extract-region in it. delete-and-extract-region is written in C. see Initializing a Variable with @code{defvar} @end ignore @node Digression into C @section Digression into C @findex delete-and-extract-region @cindex C, a digression into @cindex Digression into C The @code{copy-region-as-kill} function (@pxref{copy-region-as-kill, , @code{copy-region-as-kill}}) uses the @code{filter-buffer-substring} function, which in turn uses the @code{delete-and-extract-region} function. It removes the contents of a region and you cannot get them back. Unlike the other code discussed here, the @code{delete-and-extract-region} function is not written in Emacs Lisp; it is written in C and is one of the primitives of the GNU Emacs system. Since it is very simple, I will digress briefly from Lisp and describe it here. @need 1500 Like many of the other Emacs primitives, @code{delete-and-extract-region} is written as an instance of a C macro, a macro being a template for code. The complete macro looks like this: @c This is a copy of editfns.c's DEFUN for delete-and-extract-region. @smallexample @group DEFUN ("delete-and-extract-region", Fdelete_and_extract_region, Sdelete_and_extract_region, 2, 2, 0, doc: /* Delete the text between START and END and return it. */) (Lisp_Object start, Lisp_Object end) @{ validate_region (&start, &end); if (XFIXNUM (start) == XFIXNUM (end)) return empty_unibyte_string; return del_range_1 (XFIXNUM (start), XFIXNUM (end), 1, 1); @} @end group @end smallexample Without going into the details of the macro writing process, let me point out that this macro starts with the word @code{DEFUN}. The word @code{DEFUN} was chosen since the code serves the same purpose as @code{defun} does in Lisp. (The @code{DEFUN} C macro is defined in @file{emacs/src/lisp.h}.) The word @code{DEFUN} is followed by seven parts inside of parentheses: @itemize @bullet @item The first part is the name given to the function in Lisp, @code{delete-and-extract-region}. @item The second part is the name of the function in C, @code{Fdelete_and_extract_region}. By convention, it starts with @samp{F}. Since C does not use hyphens in names, underscores are used instead. @item The third part is the name for the C constant structure that records information on this function for internal use. It is the name of the function in C but begins with an @samp{S} instead of an @samp{F}. @item The fourth and fifth parts specify the minimum and maximum number of arguments the function can have. This function demands exactly 2 arguments. @item The sixth part is nearly like the argument that follows the @code{interactive} declaration in a function written in Lisp: a letter followed, perhaps, by a prompt. The only difference from Lisp is when the macro is called with no arguments. Then you write a @code{0} (which is a null string), as in this macro. If you were to specify arguments, you would place them between quotation marks. The C macro for @code{goto-char} includes @code{"NGoto char: "} in this position to indicate that the function expects a raw prefix, in this case, a numerical location in a buffer, and provides a prompt. @item The seventh part is a documentation string, just like the one for a function written in Emacs Lisp. This is written as a C comment. (When you build Emacs, the program @command{lib-src/make-docfile} extracts these comments and uses them to make the documentation.) @end itemize @need 1200 In a C macro, the formal parameters come next, with a statement of what kind of object they are, followed by the body of the macro. For @code{delete-and-extract-region} the body consists of the following four lines: @smallexample @group validate_region (&start, &end); if (XFIXNUM (start) == XFIXNUM (end)) return empty_unibyte_string; return del_range_1 (XFIXNUM (start), XFIXNUM (end), 1, 1); @end group @end smallexample The @code{validate_region} function checks whether the values passed as the beginning and end of the region are the proper type and are within range. If the beginning and end positions are the same, then return an empty string. The @code{del_range_1} function actually deletes the text. It is a complex function we will not look into. It updates the buffer and does other things. However, it is worth looking at the two arguments passed to @code{del_range_1}. These are @w{@code{XFIXNUM (start)}} and @w{@code{XFIXNUM (end)}}. As far as the C language is concerned, @code{start} and @code{end} are two opaque values that mark the beginning and end of the region to be deleted. More precisely, and requiring more expert knowledge to understand, the two values are of type @code{Lisp_Object}, which might be a C pointer, a C integer, or a C @code{struct}; C code ordinarily should not care how @code{Lisp_Object} is implemented. @code{Lisp_Object} widths depend on the machine, and are typically 32 or 64 bits. A few of the bits are used to specify the type of information; the remaining bits are used as content. @samp{XFIXNUM} is a C macro that extracts the relevant integer from the longer collection of bits; the type bits are discarded. @need 800 The command in @code{delete-and-extract-region} looks like this: @smallexample del_range_1 (XFIXNUM (start), XFIXNUM (end), 1, 1); @end smallexample @noindent It deletes the region between the beginning position, @code{start}, and the ending position, @code{end}. From the point of view of the person writing Lisp, Emacs is all very simple; but hidden underneath is a great deal of complexity to make it all work. @node defvar @section Initializing a Variable with @code{defvar} @findex defvar @cindex Initializing a variable @cindex Variable initialization @ignore 2006 Oct 24 In Emacs 22, copy-region-as-kill is short, 12 lines, and uses filter-buffer-substring, which is longer, 39 lines and has delete-and-extract-region in it. delete-and-extract-region is written in C. see Initializing a Variable with @code{defvar} @end ignore The @code{copy-region-as-kill} function is written in Emacs Lisp. Two functions within it, @code{kill-append} and @code{kill-new}, copy a region in a buffer and save it in a variable called the @code{kill-ring}. This section describes how the @code{kill-ring} variable is created and initialized using the @code{defvar} special form. (Again we note that the term @code{kill-ring} is a misnomer. The text that is clipped out of the buffer can be brought back; it is not a ring of corpses, but a ring of resurrectable text.) In Emacs Lisp, a variable such as the @code{kill-ring} is created and given an initial value by using the @code{defvar} special form. The name comes from ``define variable''. The @code{defvar} special form is similar to @code{setq} in that it sets the value of a variable. It is unlike @code{setq} in two ways: first, it only sets the value of the variable if the variable does not already have a value. If the variable already has a value, @code{defvar} does not override the existing value. Second, @code{defvar} has a documentation string. (There is a related macro, @code{defcustom}, designed for variables that people customize. It has more features than @code{defvar}. (@xref{defcustom, , Setting Variables with @code{defcustom}}.) @menu * See variable current value:: * defvar and asterisk:: @end menu @ifnottex @node See variable current value @unnumberedsubsec Seeing the Current Value of a Variable @end ifnottex You can see the current value of a variable, any variable, by using the @code{describe-variable} function, which is usually invoked by typing @kbd{C-h v}. If you type @kbd{C-h v} and then @code{kill-ring} (followed by @key{RET}) when prompted, you will see what is in your current kill ring---this may be quite a lot! Conversely, if you have been doing nothing this Emacs session except read this document, you may have nothing in it. Also, you will see the documentation for @code{kill-ring}: @smallexample @group Documentation: List of killed text sequences. Since the kill ring is supposed to interact nicely with cut-and-paste facilities offered by window systems, use of this variable should @end group @group interact nicely with `interprogram-cut-function' and `interprogram-paste-function'. The functions `kill-new', `kill-append', and `current-kill' are supposed to implement this interaction; you may want to use them instead of manipulating the kill ring directly. @end group @end smallexample @need 800 The kill ring is defined by a @code{defvar} in the following way: @smallexample @group (defvar kill-ring nil "List of killed text sequences. @dots{}") @end group @end smallexample @noindent In this variable definition, the variable is given an initial value of @code{nil}, which makes sense, since if you have saved nothing, you want nothing back if you give a @code{yank} command. The documentation string is written just like the documentation string of a @code{defun}. As with the documentation string of the @code{defun}, the first line of the documentation should be a complete sentence, since some commands, like @code{apropos}, print only the first line of documentation. Succeeding lines should not be indented; otherwise they look odd when you use @kbd{C-h v} (@code{describe-variable}). @node defvar and asterisk @subsection @code{defvar} and an asterisk @findex defvar @r{for a user customizable variable} @findex defvar @r{with an asterisk} In the past, Emacs used the @code{defvar} special form both for internal variables that you would not expect a user to change and for variables that you do expect a user to change. Although you can still use @code{defvar} for user customizable variables, please use @code{defcustom} instead, since it provides a path into the Customization commands. (@xref{defcustom, , Specifying Variables using @code{defcustom}}.) When you specified a variable using the @code{defvar} special form, you could distinguish a variable that a user might want to change from others by typing an asterisk, @samp{*}, in the first column of its documentation string. For example: @smallexample @group (defvar shell-command-default-error-buffer nil "*Buffer name for `shell-command' @dots{} error output. @dots{} ") @end group @end smallexample @findex set-variable @noindent You could (and still can) use the @code{set-variable} command to change the value of @code{shell-command-default-error-buffer} temporarily. However, options set using @code{set-variable} are set only for the duration of your editing session. The new values are not saved between sessions. Each time Emacs starts, it reads the original value, unless you change the value within your @file{.emacs} file, either by setting it manually or by using @code{customize}. @xref{Emacs Initialization, , Your @file{.emacs} File}. For me, the major use of the @code{set-variable} command is to suggest variables that I might want to set in my @file{.emacs} file. There are now more than 700 such variables, far too many to remember readily. Fortunately, you can press @key{TAB} after calling the @code{M-x set-variable} command to see the list of variables. (@xref{Examining, , Examining and Setting Variables, emacs, The GNU Emacs Manual}.) @need 1250 @node cons & search-fwd Review @section Review Here is a brief summary of some recently introduced functions. @table @code @item car @itemx cdr @code{car} returns the first element of a list; @code{cdr} returns the second and subsequent elements of a list. @need 1250 For example: @smallexample @group (car '(1 2 3 4 5 6 7)) @result{} 1 (cdr '(1 2 3 4 5 6 7)) @result{} (2 3 4 5 6 7) @end group @end smallexample @item cons @code{cons} constructs a list by prepending its first argument to its second argument. @need 1250 For example: @smallexample @group (cons 1 '(2 3 4)) @result{} (1 2 3 4) @end group @end smallexample @item funcall @code{funcall} evaluates its first argument as a function. It passes its remaining arguments to its first argument. @item nthcdr Return the result of taking @sc{cdr} @var{n} times on a list. @iftex The @tex $n^{th}$ @end tex @code{cdr}. @end iftex The ``rest of the rest'', as it were. @need 1250 For example: @smallexample @group (nthcdr 3 '(1 2 3 4 5 6 7)) @result{} (4 5 6 7) @end group @end smallexample @item setcar @itemx setcdr @code{setcar} changes the first element of a list; @code{setcdr} changes the second and subsequent elements of a list. @need 1250 For example: @smallexample @group (setq triple (list 1 2 3)) (setcar triple '37) triple @result{} (37 2 3) (setcdr triple '("foo" "bar")) triple @result{} (37 "foo" "bar") @end group @end smallexample @item progn Evaluate each argument in sequence and then return the value of the last. @need 1250 For example: @smallexample @group (progn 1 2 3 4) @result{} 4 @end group @end smallexample @item save-restriction Record whatever narrowing is in effect in the current buffer, if any, and restore that narrowing after evaluating the arguments. @item search-forward Search for a string, and if the string is found, move point. With a regular expression, use the similar @code{re-search-forward}. (@xref{Regexp Search, , Regular Expression Searches}, for an explanation of regular expression patterns and searches.) @need 1250 @noindent @code{search-forward} and @code{re-search-forward} take four arguments: @enumerate @item The string or regular expression to search for. @item Optionally, the limit of the search. @item Optionally, what to do if the search fails, return @code{nil} or an error message. @item Optionally, how many times to repeat the search; if negative, the search goes backwards. @end enumerate @item kill-region @itemx delete-and-extract-region @itemx copy-region-as-kill @code{kill-region} cuts the text between point and mark from the buffer and stores that text in the kill ring, so you can get it back by yanking. @code{copy-region-as-kill} copies the text between point and mark into the kill ring, from which you can get it by yanking. The function does not cut or remove the text from the buffer. @end table @code{delete-and-extract-region} removes the text between point and mark from the buffer and throws it away. You cannot get it back. (This is not an interactive command.) @need 1500 @node search Exercises @section Searching Exercises @itemize @bullet @item Write an interactive function that searches for a string. If the search finds the string, leave point after it and display a message that says ``Found!''. (Do not use @code{search-forward} for the name of this function; if you do, you will overwrite the existing version of @code{search-forward} that comes with Emacs. Use a name such as @code{test-search} instead.) @item Write a function that prints the third element of the kill ring in the echo area, if any; if the kill ring does not contain a third element, print an appropriate message. @end itemize @node List Implementation @chapter How Lists are Implemented @cindex Lists in a computer In Lisp, atoms are recorded in a straightforward fashion; if the implementation is not straightforward in practice, it is, nonetheless, straightforward in theory. The atom @samp{rose}, for example, is recorded as the four contiguous letters @samp{r}, @samp{o}, @samp{s}, @samp{e}. A list, on the other hand, is kept differently. The mechanism is equally simple, but it takes a moment to get used to the idea. A list is kept using a series of pairs of pointers. In the series, the first pointer in each pair points to an atom or to another list, and the second pointer in each pair points to the next pair, or to the symbol @code{nil}, which marks the end of the list. A pointer itself is quite simply the electronic address of what is pointed to. Hence, a list is kept as a series of electronic addresses. @menu * Lists diagrammed:: * Symbols as Chest:: Exploring a powerful metaphor. * List Exercise:: @end menu @ifnottex @node Lists diagrammed @unnumberedsec Lists diagrammed @end ifnottex For example, the list @code{(rose violet buttercup)} has three elements, @samp{rose}, @samp{violet}, and @samp{buttercup}. In the computer, the electronic address of @samp{rose} is recorded in a segment of computer memory along with the address that gives the electronic address of where the atom @samp{violet} is located; and that address (the one that tells where @samp{violet} is located) is kept along with an address that tells where the address for the atom @samp{buttercup} is located. @need 1200 This sounds more complicated than it is and is easier seen in a diagram: @c clear print-postscript-figures @c !!! cons-cell-diagram #1 @ifnottex @smallexample @group ___ ___ ___ ___ ___ ___ |___|___|--> |___|___|--> |___|___|--> nil | | | | | | --> rose --> violet --> buttercup @end group @end smallexample @end ifnottex @ifset print-postscript-figures @sp 1 @tex @center @image{cons-1} @end tex @sp 1 @end ifset @ifclear print-postscript-figures @iftex @smallexample @group ___ ___ ___ ___ ___ ___ |___|___|--> |___|___|--> |___|___|--> nil | | | | | | --> rose --> violet --> buttercup @end group @end smallexample @end iftex @end ifclear @noindent In the diagram, each box represents a word of computer memory that holds a Lisp object, usually in the form of a memory address. The boxes, i.e., the addresses, are in pairs. Each arrow points to what the address is the address of, either an atom or another pair of addresses. The first box is the electronic address of @samp{rose} and the arrow points to @samp{rose}; the second box is the address of the next pair of boxes, the first part of which is the address of @samp{violet} and the second part of which is the address of the next pair. The very last box points to the symbol @code{nil}, which marks the end of the list. @need 1200 When a variable is set to a list with an operation such as @code{setq}, it stores the address of the first box in the variable. Thus, evaluation of the expression @smallexample (setq bouquet '(rose violet buttercup)) @end smallexample @need 1250 @noindent creates a situation like this: @c cons-cell-diagram #2 @ifnottex @smallexample @group bouquet | | ___ ___ ___ ___ ___ ___ --> |___|___|--> |___|___|--> |___|___|--> nil | | | | | | --> rose --> violet --> buttercup @end group @end smallexample @end ifnottex @ifset print-postscript-figures @sp 1 @tex @center @image{cons-2} @end tex @sp 1 @end ifset @ifclear print-postscript-figures @iftex @smallexample @group bouquet | | ___ ___ ___ ___ ___ ___ --> |___|___|--> |___|___|--> |___|___|--> nil | | | | | | --> rose --> violet --> buttercup @end group @end smallexample @end iftex @end ifclear @noindent In this example, the symbol @code{bouquet} holds the address of the first pair of boxes. @need 1200 This same list can be illustrated in a different sort of box notation like this: @c cons-cell-diagram #2a @ifnottex @smallexample @group bouquet | | -------------- --------------- ---------------- | | car | cdr | | car | cdr | | car | cdr | -->| rose | o------->| violet | o------->| butter- | nil | | | | | | | | cup | | -------------- --------------- ---------------- @end group @end smallexample @end ifnottex @ifset print-postscript-figures @sp 1 @tex @center @image{cons-2a} @end tex @sp 1 @end ifset @ifclear print-postscript-figures @iftex @smallexample @group bouquet | | -------------- --------------- ---------------- | | car | cdr | | car | cdr | | car | cdr | -->| rose | o------->| violet | o------->| butter- | nil | | | | | | | | cup | | -------------- --------------- ---------------- @end group @end smallexample @end iftex @end ifclear (Symbols consist of more than pairs of addresses, but the structure of a symbol is made up of addresses. Indeed, the symbol @code{bouquet} consists of a group of address-boxes, one of which is the address of the printed word @samp{bouquet}, a second of which is the address of a function definition attached to the symbol, if any, a third of which is the address of the first pair of address-boxes for the list @code{(rose violet buttercup)}, and so on. Here we are showing that the symbol's third address-box points to the first pair of address-boxes for the list.) If a symbol is set to the @sc{cdr} of a list, the list itself is not changed; the symbol simply has an address further down the list. (In the jargon, @sc{car} and @sc{cdr} are ``non-destructive''.) Thus, evaluation of the following expression @smallexample (setq flowers (cdr bouquet)) @end smallexample @need 800 @noindent produces this: @c cons-cell-diagram #3 @ifnottex @sp 1 @smallexample @group bouquet flowers | | | ___ ___ | ___ ___ ___ ___ --> | | | --> | | | | | | |___|___|----> |___|___|--> |___|___|--> nil | | | | | | --> rose --> violet --> buttercup @end group @end smallexample @sp 1 @end ifnottex @ifset print-postscript-figures @sp 1 @tex @center @image{cons-3} @end tex @sp 1 @end ifset @ifclear print-postscript-figures @iftex @sp 1 @smallexample @group bouquet flowers | | | ___ ___ | ___ ___ ___ ___ --> | | | --> | | | | | | |___|___|----> |___|___|--> |___|___|--> nil | | | | | | --> rose --> violet --> buttercup @end group @end smallexample @sp 1 @end iftex @end ifclear @noindent The value of @code{flowers} is @code{(violet buttercup)}, which is to say, the symbol @code{flowers} holds the address of the pair of address-boxes, the first of which holds the address of @code{violet}, and the second of which holds the address of @code{buttercup}. A pair of address-boxes is called a @dfn{cons cell} or @dfn{dotted pair}. @xref{Cons Cell Type, , Cons Cell and List Types, elisp, The GNU Emacs Lisp Reference Manual}, and @ref{Dotted Pair Notation, , Dotted Pair Notation, elisp, The GNU Emacs Lisp Reference Manual}, for more information about cons cells and dotted pairs. @need 1200 The function @code{cons} adds a new pair of addresses to the front of a series of addresses like that shown above. For example, evaluating the expression @smallexample (setq bouquet (cons 'lily bouquet)) @end smallexample @need 1500 @noindent produces: @c cons-cell-diagram #4 @ifnottex @sp 1 @smallexample @group bouquet flowers | | | ___ ___ ___ ___ | ___ ___ ___ ___ --> | | | | | | --> | | | | | | |___|___|----> |___|___|----> |___|___|---->|___|___|--> nil | | | | | | | | --> lily --> rose --> violet --> buttercup @end group @end smallexample @sp 1 @end ifnottex @ifset print-postscript-figures @sp 1 @tex @center @image{cons-4} @end tex @sp 1 @end ifset @ifclear print-postscript-figures @iftex @sp 1 @smallexample @group bouquet flowers | | | ___ ___ ___ ___ | ___ ___ ___ ___ --> | | | | | | --> | | | | | | |___|___|----> |___|___|----> |___|___|---->|___|___|--> nil | | | | | | | | --> lily --> rose --> violet --> buttercup @end group @end smallexample @sp 1 @end iftex @end ifclear @need 1200 @noindent However, this does not change the value of the symbol @code{flowers}, as you can see by evaluating the following, @smallexample (eq (cdr (cdr bouquet)) flowers) @end smallexample @noindent which returns @code{t} for true. Until it is reset, @code{flowers} still has the value @code{(violet buttercup)}; that is, it has the address of the cons cell whose first address is of @code{violet}. Also, this does not alter any of the pre-existing cons cells; they are all still there. Thus, in Lisp, to get the @sc{cdr} of a list, you just get the address of the next cons cell in the series; to get the @sc{car} of a list, you get the address of the first element of the list; to @code{cons} a new element on a list, you add a new cons cell to the front of the list. That is all there is to it! The underlying structure of Lisp is brilliantly simple! And what does the last address in a series of cons cells refer to? It is the address of the empty list, of @code{nil}. In summary, when a Lisp variable is set to a value, it is provided with the address of the list to which the variable refers. @node Symbols as Chest @section Symbols as a Chest of Drawers @cindex Symbols as a Chest of Drawers @cindex Chest of Drawers, metaphor for a symbol @cindex Drawers, Chest of, metaphor for a symbol In an earlier section, I suggested that you might imagine a symbol as being a chest of drawers. The function definition is put in one drawer, the value in another, and so on. What is put in the drawer holding the value can be changed without affecting the contents of the drawer holding the function definition, and vice versa. Actually, what is put in each drawer is the address of the value or function definition. It is as if you found an old chest in the attic, and in one of its drawers you found a map giving you directions to where the buried treasure lies. (In addition to its name, symbol definition, and variable value, a symbol has a drawer for a @dfn{property list} which can be used to record other information. Property lists are not discussed here; see @ref{Property Lists, , Property Lists, elisp, The GNU Emacs Lisp Reference Manual}.) @need 1500 Here is a fanciful representation: @c chest-of-drawers diagram @ifnottex @sp 1 @smallexample @group Chest of Drawers Contents of Drawers __ o0O0o __ / \ --------------------- | directions to | [map to] | symbol name | bouquet | | +---------------------+ | directions to | | symbol definition | [none] | | +---------------------+ | directions to | [map to] | variable value | (rose violet buttercup) | | +---------------------+ | directions to | | property list | [not described here] | | +---------------------+ |/ \| @end group @end smallexample @sp 1 @end ifnottex @ifset print-postscript-figures @sp 1 @tex @center @image{drawers} @end tex @sp 1 @end ifset @ifclear print-postscript-figures @iftex @sp 1 @smallexample @group Chest of Drawers Contents of Drawers __ o0O0o __ / \ --------------------- | directions to | [map to] | symbol name | bouquet | | +---------------------+ | directions to | | symbol definition | [none] | | +---------------------+ | directions to | [map to] | variable value | (rose violet buttercup) | | +---------------------+ | directions to | | property list | [not described here] | | +---------------------+ |/ \| @end group @end smallexample @sp 1 @end iftex @end ifclear @node List Exercise @section Exercise Set @code{flowers} to @code{violet} and @code{buttercup}. Cons two more flowers on to this list and set this new list to @code{more-flowers}. Set the @sc{car} of @code{flowers} to a fish. What does the @code{more-flowers} list now contain? @node Yanking @chapter Yanking Text Back @findex yank @cindex Text retrieval @cindex Retrieving text @cindex Pasting text Whenever you cut text out of a buffer with a kill command in GNU Emacs, you can bring it back with a yank command. The text that is cut out of the buffer is put in the kill ring and the yank commands insert the appropriate contents of the kill ring back into a buffer (not necessarily the original buffer). A simple @kbd{C-y} (@code{yank}) command inserts the first item from the kill ring into the current buffer. If the @kbd{C-y} command is followed immediately by @kbd{M-y}, the first element is replaced by the second element. Successive @kbd{M-y} commands replace the second element with the third, fourth, or fifth element, and so on. When the last element in the kill ring is reached, it is replaced by the first element and the cycle is repeated. (Thus the kill ring is called a ``ring'' rather than just a ``list''. However, the actual data structure that holds the text is a list. @xref{Kill Ring, , Handling the Kill Ring}, for the details of how the list is handled as a ring.) @menu * Kill Ring Overview:: * kill-ring-yank-pointer:: The kill ring is a list. * yank nthcdr Exercises:: The @code{kill-ring-yank-pointer} variable. @end menu @node Kill Ring Overview @section Kill Ring Overview @cindex Kill ring overview The kill ring is a list of textual strings. This is what it looks like: @smallexample ("some text" "a different piece of text" "yet more text") @end smallexample If this were the contents of my kill ring and I pressed @kbd{C-y}, the string of characters saying @samp{some text} would be inserted in this buffer where my cursor is located. The @code{yank} command is also used for duplicating text by copying it. The copied text is not cut from the buffer, but a copy of it is put on the kill ring and is inserted by yanking it back. Three functions are used for bringing text back from the kill ring: @code{yank}, which is usually bound to @kbd{C-y}; @code{yank-pop}, which is usually bound to @kbd{M-y}; and @code{rotate-yank-pointer}, which is used by the two other functions. These functions refer to the kill ring through a variable called the @code{kill-ring-yank-pointer}. Indeed, the insertion code for both the @code{yank} and @code{yank-pop} functions is: @smallexample (insert (car kill-ring-yank-pointer)) @end smallexample @noindent (Well, no more. In GNU Emacs 22, the function has been replaced by @code{insert-for-yank} which calls @code{insert-for-yank-1} repetitively for each @code{yank-handler} segment. In turn, @code{insert-for-yank-1} strips text properties from the inserted text according to @code{yank-excluded-properties}. Otherwise, it is just like @code{insert}. We will stick with plain @code{insert} since it is easier to understand.) To begin to understand how @code{yank} and @code{yank-pop} work, it is first necessary to look at the @code{kill-ring-yank-pointer} variable. @node kill-ring-yank-pointer @section The @code{kill-ring-yank-pointer} Variable @code{kill-ring-yank-pointer} is a variable, just as @code{kill-ring} is a variable. It points to something by being bound to the value of what it points to, like any other Lisp variable. @need 1000 Thus, if the value of the kill ring is: @smallexample ("some text" "a different piece of text" "yet more text") @end smallexample @need 1250 @noindent and the @code{kill-ring-yank-pointer} points to the second clause, the value of @code{kill-ring-yank-pointer} is: @smallexample ("a different piece of text" "yet more text") @end smallexample As explained in the previous chapter (@pxref{List Implementation}), the computer does not keep two different copies of the text being pointed to by both the @code{kill-ring} and the @code{kill-ring-yank-pointer}. The words ``a different piece of text'' and ``yet more text'' are not duplicated. Instead, the two Lisp variables point to the same pieces of text. Here is a diagram: @c cons-cell-diagram #5 @ifnottex @smallexample @group kill-ring kill-ring-yank-pointer | | | ___ ___ | ___ ___ ___ ___ ---> | | | --> | | | | | | |___|___|----> |___|___|--> |___|___|--> nil | | | | | | | | --> "yet more text" | | | --> "a different piece of text" | --> "some text" @end group @end smallexample @sp 1 @end ifnottex @ifset print-postscript-figures @sp 1 @tex @center @image{cons-5} @end tex @sp 1 @end ifset @ifclear print-postscript-figures @iftex @smallexample @group kill-ring kill-ring-yank-pointer | | | ___ ___ | ___ ___ ___ ___ ---> | | | --> | | | | | | |___|___|----> |___|___|--> |___|___|--> nil | | | | | | | | --> "yet more text" | | | --> "a different piece of text" | --> "some text" @end group @end smallexample @sp 1 @end iftex @end ifclear Both the variable @code{kill-ring} and the variable @code{kill-ring-yank-pointer} are pointers. But the kill ring itself is usually described as if it were actually what it is composed of. The @code{kill-ring} is spoken of as if it were the list rather than that it points to the list. Conversely, the @code{kill-ring-yank-pointer} is spoken of as pointing to a list. These two ways of talking about the same thing sound confusing at first but make sense on reflection. The kill ring is generally thought of as the complete structure of data that holds the information of what has recently been cut out of the Emacs buffers. The @code{kill-ring-yank-pointer} on the other hand, serves to indicate---that is, to point to---that part of the kill ring of which the first element (the @sc{car}) will be inserted. @ignore In GNU Emacs 22, the @code{kill-new} function calls @code{(setq kill-ring-yank-pointer kill-ring)} (defun rotate-yank-pointer (arg) "Rotate the yanking point in the kill ring. With argument, rotate that many kills forward (or backward, if negative)." (interactive "p") (current-kill arg)) (defun current-kill (n &optional do-not-move) "Rotate the yanking point by N places, and then return that kill. If N is zero, `interprogram-paste-function' is set, and calling it returns a string, then that string is added to the front of the kill ring and returned as the latest kill. If optional arg DO-NOT-MOVE is non-nil, then don't actually move the yanking point; just return the Nth kill forward." (let ((interprogram-paste (and (= n 0) interprogram-paste-function (funcall interprogram-paste-function)))) (if interprogram-paste (progn ;; Disable the interprogram cut function when we add the new ;; text to the kill ring, so Emacs doesn't try to own the ;; selection, with identical text. (let ((interprogram-cut-function nil)) (kill-new interprogram-paste)) interprogram-paste) (or kill-ring (error "Kill ring is empty")) (let ((ARGth-kill-element (nthcdr (mod (- n (length kill-ring-yank-pointer)) (length kill-ring)) kill-ring))) (or do-not-move (setq kill-ring-yank-pointer ARGth-kill-element)) (car ARGth-kill-element))))) @end ignore @need 1500 @node yank nthcdr Exercises @section Exercises with @code{yank} and @code{nthcdr} @itemize @bullet @item Using @kbd{C-h v} (@code{describe-variable}), look at the value of your kill ring. Add several items to your kill ring; look at its value again. Using @kbd{M-y} (@code{yank-pop)}, move all the way around the kill ring. How many items were in your kill ring? Find the value of @code{kill-ring-max}. Was your kill ring full, or could you have kept more blocks of text within it? @item Using @code{nthcdr} and @code{car}, construct a series of expressions to return the first, second, third, and fourth elements of a list. @end itemize @node Loops & Recursion @chapter Loops and Recursion @cindex Loops and recursion @cindex Recursion and loops @cindex Repetition (loops) Emacs Lisp has two primary ways to cause an expression, or a series of expressions, to be evaluated repeatedly: one uses a @code{while} loop, and the other uses @dfn{recursion}. Repetition can be very valuable. For example, to move forward four sentences, you need only write a program that will move forward one sentence and then repeat the process four times. Since a computer does not get bored or tired, such repetitive action does not have the deleterious effects that excessive or the wrong kinds of repetition can have on humans. People mostly write Emacs Lisp functions using @code{while} loops and their kin; but you can use recursion, which provides a very powerful way to think about and then to solve problems@footnote{You can write recursive functions to be frugal or wasteful of mental or computer resources; as it happens, methods that people find easy---that are frugal of mental resources---sometimes use considerable computer resources. Emacs was designed to run on machines that we now consider limited and its default settings are conservative. You may want to increase the values of @code{max-specpdl-size} and @code{max-lisp-eval-depth}. In my @file{.emacs} file, I set them to 15 and 30 times their default value.}. @menu * while:: Causing a stretch of code to repeat. * dolist dotimes:: * Recursion:: Causing a function to call itself. * Looping exercise:: @end menu @node while @section @code{while} @cindex Loops @findex while The @code{while} special form tests whether the value returned by evaluating its first argument is true or false. This is similar to what the Lisp interpreter does with an @code{if}; what the interpreter does next, however, is different. In a @code{while} expression, if the value returned by evaluating the first argument is false, the Lisp interpreter skips the rest of the expression (the @dfn{body} of the expression) and does not evaluate it. However, if the value is true, the Lisp interpreter evaluates the body of the expression and then again tests whether the first argument to @code{while} is true or false. If the value returned by evaluating the first argument is again true, the Lisp interpreter again evaluates the body of the expression. @need 1200 The template for a @code{while} expression looks like this: @smallexample @group (while @var{true-or-false-test} @var{body}@dots{}) @end group @end smallexample @menu * Looping with while:: Repeat so long as test returns true. * Loop Example:: A @code{while} loop that uses a list. * print-elements-of-list:: Uses @code{while}, @code{car}, @code{cdr}. * Incrementing Loop:: A loop with an incrementing counter. * Incrementing Loop Details:: * Decrementing Loop:: A loop with a decrementing counter. @end menu @ifnottex @node Looping with while @unnumberedsubsec Looping with @code{while} @end ifnottex So long as the true-or-false-test of the @code{while} expression returns a true value when it is evaluated, the body is repeatedly evaluated. This process is called a loop since the Lisp interpreter repeats the same thing again and again, like an airplane doing a loop. When the result of evaluating the true-or-false-test is false, the Lisp interpreter does not evaluate the rest of the @code{while} expression and exits the loop. Clearly, if the value returned by evaluating the first argument to @code{while} is always true, the body following will be evaluated again and again @dots{} and again @dots{} forever. Conversely, if the value returned is never true, the expressions in the body will never be evaluated. The craft of writing a @code{while} loop consists of choosing a mechanism such that the true-or-false-test returns true just the number of times that you want the subsequent expressions to be evaluated, and then have the test return false. The value returned by evaluating a @code{while} is the value of the true-or-false-test. An interesting consequence of this is that a @code{while} loop that evaluates without error will return @code{nil} or false regardless of whether it has looped 1 or 100 times or none at all. A @code{while} expression that evaluates successfully never returns a true value! What this means is that @code{while} is always evaluated for its side effects, which is to say, the consequences of evaluating the expressions within the body of the @code{while} loop. This makes sense. It is not the mere act of looping that is desired, but the consequences of what happens when the expressions in the loop are repeatedly evaluated. @node Loop Example @subsection A @code{while} Loop and a List A common way to control a @code{while} loop is to test whether a list has any elements. If it does, the loop is repeated; but if it does not, the repetition is ended. Since this is an important technique, we will create a short example to illustrate it. A simple way to test whether a list has elements is to evaluate the list: if it has no elements, it is an empty list and will return the empty list, @code{()}, which is a synonym for @code{nil} or false. On the other hand, a list with elements will return those elements when it is evaluated. Since Emacs Lisp considers as true any value that is not @code{nil}, a list that returns elements will test true in a @code{while} loop. @need 1200 For example, you can set the variable @code{empty-list} to @code{nil} by evaluating the following @code{setq} expression: @smallexample (setq empty-list ()) @end smallexample @noindent After evaluating the @code{setq} expression, you can evaluate the variable @code{empty-list} in the usual way, by placing the cursor after the symbol and typing @kbd{C-x C-e}; @code{nil} will appear in your echo area: @smallexample empty-list @end smallexample On the other hand, if you set a variable to be a list with elements, the list will appear when you evaluate the variable, as you can see by evaluating the following two expressions: @smallexample @group (setq animals '(gazelle giraffe lion tiger)) animals @end group @end smallexample Thus, to create a @code{while} loop that tests whether there are any items in the list @code{animals}, the first part of the loop will be written like this: @smallexample @group (while animals @dots{} @end group @end smallexample @noindent When the @code{while} tests its first argument, the variable @code{animals} is evaluated. It returns a list. So long as the list has elements, the @code{while} considers the results of the test to be true; but when the list is empty, it considers the results of the test to be false. To prevent the @code{while} loop from running forever, some mechanism needs to be provided to empty the list eventually. An oft-used technique is to have one of the subsequent forms in the @code{while} expression set the value of the list to be the @sc{cdr} of the list. Each time the @code{cdr} function is evaluated, the list will be made shorter, until eventually only the empty list will be left. At this point, the test of the @code{while} loop will return false, and the arguments to the @code{while} will no longer be evaluated. For example, the list of animals bound to the variable @code{animals} can be set to be the @sc{cdr} of the original list with the following expression: @smallexample (setq animals (cdr animals)) @end smallexample @noindent If you have evaluated the previous expressions and then evaluate this expression, you will see @code{(giraffe lion tiger)} appear in the echo area. If you evaluate the expression again, @code{(lion tiger)} will appear in the echo area. If you evaluate it again and yet again, @code{(tiger)} appears and then the empty list, shown by @code{nil}. A template for a @code{while} loop that uses the @code{cdr} function repeatedly to cause the true-or-false-test eventually to test false looks like this: @smallexample @group (while @var{test-whether-list-is-empty} @var{body}@dots{} @var{set-list-to-cdr-of-list}) @end group @end smallexample This test and use of @code{cdr} can be put together in a function that goes through a list and prints each element of the list on a line of its own. @node print-elements-of-list @subsection An Example: @code{print-elements-of-list} @findex print-elements-of-list The @code{print-elements-of-list} function illustrates a @code{while} loop with a list. @cindex @file{*scratch*} buffer The function requires several lines for its output. If you are reading this in a recent instance of GNU Emacs, @c GNU Emacs 21, GNU Emacs 22, or a later version, you can evaluate the following expression inside of Info, as usual. If you are using an earlier version of Emacs, you need to copy the necessary expressions to your @file{*scratch*} buffer and evaluate them there. This is because the echo area had only one line in the earlier versions. You can copy the expressions by marking the beginning of the region with @kbd{C-@key{SPC}} (@code{set-mark-command}), moving the cursor to the end of the region and then copying the region using @kbd{M-w} (@code{kill-ring-save}, which calls @code{copy-region-as-kill} and then provides visual feedback). In the @file{*scratch*} buffer, you can yank the expressions back by typing @kbd{C-y} (@code{yank}). After you have copied the expressions to the @file{*scratch*} buffer, evaluate each expression in turn. Be sure to evaluate the last expression, @code{(print-elements-of-list animals)}, by typing @kbd{C-u C-x C-e}, that is, by giving an argument to @code{eval-last-sexp}. This will cause the result of the evaluation to be printed in the @file{*scratch*} buffer instead of being printed in the echo area. (Otherwise you will see something like this in your echo area: @code{^Jgazelle^J^Jgiraffe^J^Jlion^J^Jtiger^Jnil}, in which each @samp{^J} stands for a newline.) @need 1500 In a recent instance of GNU Emacs, you can evaluate these expressions directly in the Info buffer, and the echo area will grow to show the results. @smallexample @group (setq animals '(gazelle giraffe lion tiger)) (defun print-elements-of-list (list) "Print each element of LIST on a line of its own." (while list (print (car list)) (setq list (cdr list)))) (print-elements-of-list animals) @end group @end smallexample @need 1200 @noindent When you evaluate the three expressions in sequence, you will see this: @smallexample @group gazelle giraffe lion tiger nil @end group @end smallexample Each element of the list is printed on a line of its own (that is what the function @code{print} does) and then the value returned by the function is printed. Since the last expression in the function is the @code{while} loop, and since @code{while} loops always return @code{nil}, a @code{nil} is printed after the last element of the list. @node Incrementing Loop @subsection A Loop with an Incrementing Counter A loop is not useful unless it stops when it ought. Besides controlling a loop with a list, a common way of stopping a loop is to write the first argument as a test that returns false when the correct number of repetitions are complete. This means that the loop must have a counter---an expression that counts how many times the loop repeats itself. @ifnottex @node Incrementing Loop Details @unnumberedsubsec Details of an Incrementing Loop @end ifnottex The test for a loop with an incrementing counter can be an expression such as @code{(< count desired-number)} which returns @code{t} for true if the value of @code{count} is less than the @code{desired-number} of repetitions and @code{nil} for false if the value of @code{count} is equal to or is greater than the @code{desired-number}. The expression that increments the count can be a simple @code{setq} such as @code{(setq count (1+ count))}, where @code{1+} is a built-in function in Emacs Lisp that adds 1 to its argument. (The expression @w{@code{(1+ count)}} has the same result as @w{@code{(+ count 1)}}, but is easier for a human to read.) @need 1250 The template for a @code{while} loop controlled by an incrementing counter looks like this: @smallexample @group @var{set-count-to-initial-value} (while (< count desired-number) ; @r{true-or-false-test} @var{body}@dots{} (setq count (1+ count))) ; @r{incrementer} @end group @end smallexample @noindent Note that you need to set the initial value of @code{count}; usually it is set to 1. @menu * Incrementing Example:: Counting pebbles in a triangle. * Inc Example parts:: The parts of the function definition. * Inc Example altogether:: Putting the function definition together. @end menu @node Incrementing Example @unnumberedsubsubsec Example with incrementing counter Suppose you are playing on the beach and decide to make a triangle of pebbles, putting one pebble in the first row, two in the second row, three in the third row and so on, like this: @sp 1 @c pebble diagram @ifnottex @smallexample @group * * * * * * * * * * @end group @end smallexample @end ifnottex @iftex @smallexample @group @bullet{} @bullet{} @bullet{} @bullet{} @bullet{} @bullet{} @bullet{} @bullet{} @bullet{} @bullet{} @end group @end smallexample @end iftex @sp 1 @noindent (About 2500 years ago, Pythagoras and others developed the beginnings of number theory by considering questions such as this.) Suppose you want to know how many pebbles you will need to make a triangle with 7 rows? Clearly, what you need to do is add up the numbers from 1 to 7. There are two ways to do this; start with the smallest number, one, and add up the list in sequence, 1, 2, 3, 4 and so on; or start with the largest number and add the list going down: 7, 6, 5, 4 and so on. Because both mechanisms illustrate common ways of writing @code{while} loops, we will create two examples, one counting up and the other counting down. In this first example, we will start with 1 and add 2, 3, 4 and so on. If you are just adding up a short list of numbers, the easiest way to do it is to add up all the numbers at once. However, if you do not know ahead of time how many numbers your list will have, or if you want to be prepared for a very long list, then you need to design your addition so that what you do is repeat a simple process many times instead of doing a more complex process once. For example, instead of adding up all the pebbles all at once, what you can do is add the number of pebbles in the first row, 1, to the number in the second row, 2, and then add the total of those two rows to the third row, 3. Then you can add the number in the fourth row, 4, to the total of the first three rows; and so on. The critical characteristic of the process is that each repetitive action is simple. In this case, at each step we add only two numbers, the number of pebbles in the row and the total already found. This process of adding two numbers is repeated again and again until the last row has been added to the total of all the preceding rows. In a more complex loop the repetitive action might not be so simple, but it will be simpler than doing everything all at once. @node Inc Example parts @unnumberedsubsubsec The parts of the function definition The preceding analysis gives us the bones of our function definition: first, we will need a variable that we can call @code{total} that will be the total number of pebbles. This will be the value returned by the function. Second, we know that the function will require an argument: this argument will be the total number of rows in the triangle. It can be called @code{number-of-rows}. Finally, we need a variable to use as a counter. We could call this variable @code{counter}, but a better name is @code{row-number}. That is because what the counter does in this function is count rows, and a program should be written to be as understandable as possible. When the Lisp interpreter first starts evaluating the expressions in the function, the value of @code{total} should be set to zero, since we have not added anything to it. Then the function should add the number of pebbles in the first row to the total, and then add the number of pebbles in the second to the total, and then add the number of pebbles in the third row to the total, and so on, until there are no more rows left to add. Both @code{total} and @code{row-number} are used only inside the function, so they can be declared as local variables with @code{let} and given initial values. Clearly, the initial value for @code{total} should be 0. The initial value of @code{row-number} should be 1, since we start with the first row. This means that the @code{let} statement will look like this: @smallexample @group (let ((total 0) (row-number 1)) @var{body}@dots{}) @end group @end smallexample After the internal variables are declared and bound to their initial values, we can begin the @code{while} loop. The expression that serves as the test should return a value of @code{t} for true so long as the @code{row-number} is less than or equal to the @code{number-of-rows}. (If the expression tests true only so long as the row number is less than the number of rows in the triangle, the last row will never be added to the total; hence the row number has to be either less than or equal to the number of rows.) @need 1500 @findex <= @r{(less than or equal)} Lisp provides the @code{<=} function that returns true if the value of its first argument is less than or equal to the value of its second argument and false otherwise. So the expression that the @code{while} will evaluate as its test should look like this: @smallexample (<= row-number number-of-rows) @end smallexample The total number of pebbles can be found by repeatedly adding the number of pebbles in a row to the total already found. Since the number of pebbles in the row is equal to the row number, the total can be found by adding the row number to the total. (Clearly, in a more complex situation, the number of pebbles in the row might be related to the row number in a more complicated way; if this were the case, the row number would be replaced by the appropriate expression.) @smallexample (setq total (+ total row-number)) @end smallexample @noindent What this does is set the new value of @code{total} to be equal to the sum of adding the number of pebbles in the row to the previous total. After setting the value of @code{total}, the conditions need to be established for the next repetition of the loop, if there is one. This is done by incrementing the value of the @code{row-number} variable, which serves as a counter. After the @code{row-number} variable has been incremented, the true-or-false-test at the beginning of the @code{while} loop tests whether its value is still less than or equal to the value of the @code{number-of-rows} and if it is, adds the new value of the @code{row-number} variable to the @code{total} of the previous repetition of the loop. @need 1200 The built-in Emacs Lisp function @code{1+} adds 1 to a number, so the @code{row-number} variable can be incremented with this expression: @smallexample (setq row-number (1+ row-number)) @end smallexample @node Inc Example altogether @unnumberedsubsubsec Putting the function definition together We have created the parts for the function definition; now we need to put them together. @need 800 First, the contents of the @code{while} expression: @smallexample @group (while (<= row-number number-of-rows) ; @r{true-or-false-test} (setq total (+ total row-number)) (setq row-number (1+ row-number))) ; @r{incrementer} @end group @end smallexample Along with the @code{let} expression varlist, this very nearly completes the body of the function definition. However, it requires one final element, the need for which is somewhat subtle. The final touch is to place the variable @code{total} on a line by itself after the @code{while} expression. Otherwise, the value returned by the whole function is the value of the last expression that is evaluated in the body of the @code{let}, and this is the value returned by the @code{while}, which is always @code{nil}. This may not be evident at first sight. It almost looks as if the incrementing expression is the last expression of the whole function. But that expression is part of the body of the @code{while}; it is the last element of the list that starts with the symbol @code{while}. Moreover, the whole of the @code{while} loop is a list within the body of the @code{let}. @need 1250 In outline, the function will look like this: @smallexample @group (defun @var{name-of-function} (@var{argument-list}) "@var{documentation}@dots{}" (let (@var{varlist}) (while (@var{true-or-false-test}) @var{body-of-while}@dots{} ) @dots{} )) ; @r{Need final expression here.} @end group @end smallexample The result of evaluating the @code{let} is what is going to be returned by the @code{defun} since the @code{let} is not embedded within any containing list, except for the @code{defun} as a whole. However, if the @code{while} is the last element of the @code{let} expression, the function will always return @code{nil}. This is not what we want! Instead, what we want is the value of the variable @code{total}. This is returned by simply placing the symbol as the last element of the list starting with @code{let}. It gets evaluated after the preceding elements of the list are evaluated, which means it gets evaluated after it has been assigned the correct value for the total. It may be easier to see this by printing the list starting with @code{let} all on one line. This format makes it evident that the @var{varlist} and @code{while} expressions are the second and third elements of the list starting with @code{let}, and the @code{total} is the last element: @smallexample @group (let (@var{varlist}) (while (@var{true-or-false-test}) @var{body-of-while}@dots{} ) total) @end group @end smallexample @need 1200 Putting everything together, the @code{triangle} function definition looks like this: @smallexample @group (defun triangle (number-of-rows) ; @r{Version with} ; @r{ incrementing counter.} "Add up the number of pebbles in a triangle. The first row has one pebble, the second row two pebbles, the third row three pebbles, and so on. The argument is NUMBER-OF-ROWS." @end group @group (let ((total 0) (row-number 1)) (while (<= row-number number-of-rows) (setq total (+ total row-number)) (setq row-number (1+ row-number))) total)) @end group @end smallexample @need 1200 After you have installed @code{triangle} by evaluating the function, you can try it out. Here are two examples: @smallexample @group (triangle 4) (triangle 7) @end group @end smallexample @noindent The sum of the first four numbers is 10 and the sum of the first seven numbers is 28. @node Decrementing Loop @subsection Loop with a Decrementing Counter Another common way to write a @code{while} loop is to write the test so that it determines whether a counter is greater than zero. So long as the counter is greater than zero, the loop is repeated. But when the counter is equal to or less than zero, the loop is stopped. For this to work, the counter has to start out greater than zero and then be made smaller and smaller by a form that is evaluated repeatedly. The test will be an expression such as @code{(> counter 0)} which returns @code{t} for true if the value of @code{counter} is greater than zero, and @code{nil} for false if the value of @code{counter} is equal to or less than zero. The expression that makes the number smaller and smaller can be a simple @code{setq} such as @code{(setq counter (1- counter))}, where @code{1-} is a built-in function in Emacs Lisp that subtracts 1 from its argument. @need 1250 The template for a decrementing @code{while} loop looks like this: @smallexample @group (while (> counter 0) ; @r{true-or-false-test} @var{body}@dots{} (setq counter (1- counter))) ; @r{decrementer} @end group @end smallexample @menu * Decrementing Example:: More pebbles on the beach. * Dec Example parts:: The parts of the function definition. * Dec Example altogether:: Putting the function definition together. @end menu @node Decrementing Example @unnumberedsubsubsec Example with decrementing counter To illustrate a loop with a decrementing counter, we will rewrite the @code{triangle} function so the counter decreases to zero. This is the reverse of the earlier version of the function. In this case, to find out how many pebbles are needed to make a triangle with 3 rows, add the number of pebbles in the third row, 3, to the number in the preceding row, 2, and then add the total of those two rows to the row that precedes them, which is 1. Likewise, to find the number of pebbles in a triangle with 7 rows, add the number of pebbles in the seventh row, 7, to the number in the preceding row, which is 6, and then add the total of those two rows to the row that precedes them, which is 5, and so on. As in the previous example, each addition only involves adding two numbers, the total of the rows already added up and the number of pebbles in the row that is being added to the total. This process of adding two numbers is repeated again and again until there are no more pebbles to add. We know how many pebbles to start with: the number of pebbles in the last row is equal to the number of rows. If the triangle has seven rows, the number of pebbles in the last row is 7. Likewise, we know how many pebbles are in the preceding row: it is one less than the number in the row. @node Dec Example parts @unnumberedsubsubsec The parts of the function definition We start with three variables: the total number of rows in the triangle; the number of pebbles in a row; and the total number of pebbles, which is what we want to calculate. These variables can be named @code{number-of-rows}, @code{number-of-pebbles-in-row}, and @code{total}, respectively. Both @code{total} and @code{number-of-pebbles-in-row} are used only inside the function and are declared with @code{let}. The initial value of @code{total} should, of course, be zero. However, the initial value of @code{number-of-pebbles-in-row} should be equal to the number of rows in the triangle, since the addition will start with the longest row. @need 1250 This means that the beginning of the @code{let} expression will look like this: @smallexample @group (let ((total 0) (number-of-pebbles-in-row number-of-rows)) @var{body}@dots{}) @end group @end smallexample The total number of pebbles can be found by repeatedly adding the number of pebbles in a row to the total already found, that is, by repeatedly evaluating the following expression: @smallexample (setq total (+ total number-of-pebbles-in-row)) @end smallexample @noindent After the @code{number-of-pebbles-in-row} is added to the @code{total}, the @code{number-of-pebbles-in-row} should be decremented by one, since the next time the loop repeats, the preceding row will be added to the total. The number of pebbles in a preceding row is one less than the number of pebbles in a row, so the built-in Emacs Lisp function @code{1-} can be used to compute the number of pebbles in the preceding row. This can be done with the following expression: @smallexample @group (setq number-of-pebbles-in-row (1- number-of-pebbles-in-row)) @end group @end smallexample Finally, we know that the @code{while} loop should stop making repeated additions when there are no pebbles in a row. So the test for the @code{while} loop is simply: @smallexample (while (> number-of-pebbles-in-row 0) @end smallexample @node Dec Example altogether @unnumberedsubsubsec Putting the function definition together We can put these expressions together to create a function definition that works. However, on examination, we find that one of the local variables is unneeded! @need 1250 The function definition looks like this: @smallexample @group ;;; @r{First subtractive version.} (defun triangle (number-of-rows) "Add up the number of pebbles in a triangle." (let ((total 0) (number-of-pebbles-in-row number-of-rows)) (while (> number-of-pebbles-in-row 0) (setq total (+ total number-of-pebbles-in-row)) (setq number-of-pebbles-in-row (1- number-of-pebbles-in-row))) total)) @end group @end smallexample As written, this function works. However, we do not need @code{number-of-pebbles-in-row}. @cindex Argument as local variable When the @code{triangle} function is evaluated, the symbol @code{number-of-rows} will be bound to a number, giving it an initial value. That number can be changed in the body of the function as if it were a local variable, without any fear that such a change will effect the value of the variable outside of the function. This is a very useful characteristic of Lisp; it means that the variable @code{number-of-rows} can be used anywhere in the function where @code{number-of-pebbles-in-row} is used. @need 800 Here is a second version of the function written a bit more cleanly: @smallexample @group (defun triangle (number) ; @r{Second version.} "Return sum of numbers 1 through NUMBER inclusive." (let ((total 0)) (while (> number 0) (setq total (+ total number)) (setq number (1- number))) total)) @end group @end smallexample In brief, a properly written @code{while} loop will consist of three parts: @enumerate @item A test that will return false after the loop has repeated itself the correct number of times. @item An expression the evaluation of which will return the value desired after being repeatedly evaluated. @item An expression to change the value passed to the true-or-false-test so that the test returns false after the loop has repeated itself the right number of times. @end enumerate @node dolist dotimes @section Save your time: @code{dolist} and @code{dotimes} In addition to @code{while}, both @code{dolist} and @code{dotimes} provide for looping. Sometimes these are quicker to write than the equivalent @code{while} loop. Both are Lisp macros. (@xref{Macros, , Macros, elisp, The GNU Emacs Lisp Reference Manual}. ) @code{dolist} works like a @code{while} loop that @sc{cdr}s down a list: @code{dolist} automatically shortens the list each time it loops---takes the @sc{cdr} of the list---and binds the @sc{car} of each shorter version of the list to the first of its arguments. @code{dotimes} loops a specific number of times: you specify the number. @menu * dolist:: * dotimes:: @end menu @node dolist @unnumberedsubsec The @code{dolist} Macro @findex dolist Suppose, for example, you want to reverse a list, so that ``first'' ``second'' ``third'' becomes ``third'' ``second'' ``first''. @need 1250 In practice, you would use the @code{reverse} function, like this: @smallexample @group (setq animals '(gazelle giraffe lion tiger)) (reverse animals) @end group @end smallexample @need 800 @noindent Here is how you could reverse the list using a @code{while} loop: @smallexample @group (setq animals '(gazelle giraffe lion tiger)) (defun reverse-list-with-while (list) "Using while, reverse the order of LIST." (let (value) ; make sure list starts empty (while list (setq value (cons (car list) value)) (setq list (cdr list))) value)) (reverse-list-with-while animals) @end group @end smallexample @need 800 @noindent And here is how you could use the @code{dolist} macro: @smallexample @group (setq animals '(gazelle giraffe lion tiger)) (defun reverse-list-with-dolist (list) "Using dolist, reverse the order of LIST." (let (value) ; make sure list starts empty (dolist (element list value) (setq value (cons element value))))) (reverse-list-with-dolist animals) @end group @end smallexample @need 1250 @noindent In Info, you can place your cursor after the closing parenthesis of each expression and type @kbd{C-x C-e}; in each case, you should see @smallexample (tiger lion giraffe gazelle) @end smallexample @noindent in the echo area. For this example, the existing @code{reverse} function is obviously best. The @code{while} loop is just like our first example (@pxref{Loop Example, , A @code{while} Loop and a List}). The @code{while} first checks whether the list has elements; if so, it constructs a new list by adding the first element of the list to the existing list (which in the first iteration of the loop is @code{nil}). Since the second element is prepended in front of the first element, and the third element is prepended in front of the second element, the list is reversed. In the expression using a @code{while} loop, the @w{@code{(setq list (cdr list))}} expression shortens the list, so the @code{while} loop eventually stops. In addition, it provides the @code{cons} expression with a new first element by creating a new and shorter list at each repetition of the loop. The @code{dolist} expression does very much the same as the @code{while} expression, except that the @code{dolist} macro does some of the work you have to do when writing a @code{while} expression. Like a @code{while} loop, a @code{dolist} loops. What is different is that it automatically shortens the list each time it loops---it @sc{cdr}s down the list on its own---and it automatically binds the @sc{car} of each shorter version of the list to the first of its arguments. In the example, the @sc{car} of each shorter version of the list is referred to using the symbol @samp{element}, the list itself is called @samp{list}, and the value returned is called @samp{value}. The remainder of the @code{dolist} expression is the body. The @code{dolist} expression binds the @sc{car} of each shorter version of the list to @code{element} and then evaluates the body of the expression; and repeats the loop. The result is returned in @code{value}. @node dotimes @unnumberedsubsec The @code{dotimes} Macro @findex dotimes The @code{dotimes} macro is similar to @code{dolist}, except that it loops a specific number of times. The first argument to @code{dotimes} is assigned the numbers 0, 1, 2 and so forth each time around the loop. You need to provide the value of the second argument, which is how many times the macro loops. @need 1250 For example, the following binds the numbers from 0 up to, but not including, the number 3 to the first argument, @var{number}, and then constructs a list of the three numbers. (The first number is 0, the second number is 1, and the third number is 2; this makes a total of three numbers in all, starting with zero as the first number.) @smallexample @group (let (value) ; otherwise a value is a void variable (dotimes (number 3) (setq value (cons number value))) value) @result{} (2 1 0) @end group @end smallexample @noindent The way to use @code{dotimes} is to operate on some expression @var{number} number of times and then return the result, either as a list or an atom. @need 1250 Here is an example of a @code{defun} that uses @code{dotimes} to add up the number of pebbles in a triangle. @smallexample @group (defun triangle-using-dotimes (number-of-rows) "Using `dotimes', add up the number of pebbles in a triangle." (let ((total 0)) ; otherwise a total is a void variable (dotimes (number number-of-rows) (setq total (+ total (1+ number)))) total)) (triangle-using-dotimes 4) @end group @end smallexample @node Recursion @section Recursion @cindex Recursion A recursive function contains code that tells the Lisp interpreter to call a program that runs exactly like itself, but with slightly different arguments. The code runs exactly the same because it has the same name. However, even though the program has the same name, it is not the same entity. It is different. In the jargon, it is a different ``instance''. Eventually, if the program is written correctly, the slightly different arguments will become sufficiently different from the first arguments that the final instance will stop. @menu * Building Robots:: Same model, different serial number ... * Recursive Definition Parts:: Walk until you stop ... * Recursion with list:: Using a list as the test whether to recurse. * Recursive triangle function:: * Recursion with cond:: * Recursive Patterns:: Often used templates. * No Deferment:: Don't store up work ... * No deferment solution:: @end menu @node Building Robots @subsection Building Robots: Extending the Metaphor @cindex Building robots @cindex Robots, building It is sometimes helpful to think of a running program as a robot that does a job. In doing its job, a recursive function calls on a second robot to help it. The second robot is identical to the first in every way, except that the second robot helps the first and has been passed different arguments than the first. In a recursive function, the second robot may call a third; and the third may call a fourth, and so on. Each of these is a different entity; but all are clones. Since each robot has slightly different instructions---the arguments will differ from one robot to the next---the last robot should know when to stop. Let's expand on the metaphor in which a computer program is a robot. A function definition provides the blueprints for a robot. When you install a function definition, that is, when you evaluate a @code{defun} macro, you install the necessary equipment to build robots. It is as if you were in a factory, setting up an assembly line. Robots with the same name are built according to the same blueprints. So they have the same model number, but a different serial number. We often say that a recursive function ``calls itself''. What we mean is that the instructions in a recursive function cause the Lisp interpreter to run a different function that has the same name and does the same job as the first, but with different arguments. It is important that the arguments differ from one instance to the next; otherwise, the process will never stop. @node Recursive Definition Parts @subsection The Parts of a Recursive Definition @cindex Parts of a Recursive Definition @cindex Recursive Definition Parts A recursive function typically contains a conditional expression which has three parts: @enumerate @item A true-or-false-test that determines whether the function is called again, here called the @dfn{do-again-test}. @item The name of the function. When this name is called, a new instance of the function---a new robot, as it were---is created and told what to do. @item An expression that returns a different value each time the function is called, here called the @dfn{next-step-expression}. Consequently, the argument (or arguments) passed to the new instance of the function will be different from that passed to the previous instance. This causes the conditional expression, the @dfn{do-again-test}, to test false after the correct number of repetitions. @end enumerate Recursive functions can be much simpler than any other kind of function. Indeed, when people first start to use them, they often look so mysteriously simple as to be incomprehensible. Like riding a bicycle, reading a recursive function definition takes a certain knack which is hard at first but then seems simple. @need 1200 There are several different common recursive patterns. A very simple pattern looks like this: @smallexample @group (defun @var{name-of-recursive-function} (@var{argument-list}) "@var{documentation}@dots{}" (if @var{do-again-test} @var{body}@dots{} (@var{name-of-recursive-function} @var{next-step-expression}))) @end group @end smallexample Each time a recursive function is evaluated, a new instance of it is created and told what to do. The arguments tell the instance what to do. An argument is bound to the value of the next-step-expression. Each instance runs with a different value of the next-step-expression. The value in the next-step-expression is used in the do-again-test. The value returned by the next-step-expression is passed to the new instance of the function, which evaluates it (or some transmogrification of it) to determine whether to continue or stop. The next-step-expression is designed so that the do-again-test returns false when the function should no longer be repeated. The do-again-test is sometimes called the @dfn{stop condition}, since it stops the repetitions when it tests false. @node Recursion with list @subsection Recursion with a List The example of a @code{while} loop that printed the elements of a list of numbers can be written recursively. Here is the code, including an expression to set the value of the variable @code{animals} to a list. If you are reading this in Info in Emacs, you can evaluate this expression directly in Info. Otherwise, you must copy the example to the @file{*scratch*} buffer and evaluate each expression there. Use @kbd{C-u C-x C-e} to evaluate the @code{(print-elements-recursively animals)} expression so that the results are printed in the buffer; otherwise the Lisp interpreter will try to squeeze the results into the one line of the echo area. Also, place your cursor immediately after the last closing parenthesis of the @code{print-elements-recursively} function, before the comment. Otherwise, the Lisp interpreter will try to evaluate the comment. @findex print-elements-recursively @smallexample @group (setq animals '(gazelle giraffe lion tiger)) (defun print-elements-recursively (list) "Print each element of LIST on a line of its own. Uses recursion." (when list ; @r{do-again-test} (print (car list)) ; @r{body} (print-elements-recursively ; @r{recursive call} (cdr list)))) ; @r{next-step-expression} (print-elements-recursively animals) @end group @end smallexample The @code{print-elements-recursively} function first tests whether there is any content in the list; if there is, the function prints the first element of the list, the @sc{car} of the list. Then the function invokes itself, but gives itself as its argument, not the whole list, but the second and subsequent elements of the list, the @sc{cdr} of the list. Put another way, if the list is not empty, the function invokes another instance of code that is similar to the initial code, but is a different thread of execution, with different arguments than the first instance. Put in yet another way, if the list is not empty, the first robot assembles a second robot and tells it what to do; the second robot is a different individual from the first, but is the same model. When the second evaluation occurs, the @code{when} expression is evaluated and if true, prints the first element of the list it receives as its argument (which is the second element of the original list). Then the function calls itself with the @sc{cdr} of the list it is invoked with, which (the second time around) is the @sc{cdr} of the @sc{cdr} of the original list. Note that although we say that the function ``calls itself'', what we mean is that the Lisp interpreter assembles and instructs a new instance of the program. The new instance is a clone of the first, but is a separate individual. Each time the function invokes itself, it does so on a shorter version of the original list. It creates a new instance that works on a shorter list. Eventually, the function invokes itself on an empty list. It creates a new instance whose argument is @code{nil}. The conditional expression tests the value of @code{list}. Since the value of @code{list} is @code{nil}, the @code{when} expression tests false so the then-part is not evaluated. The function as a whole then returns @code{nil}. @need 1200 When you evaluate the expression @code{(print-elements-recursively animals)} in the @file{*scratch*} buffer, you see this result: @smallexample @group gazelle giraffe lion tiger nil @end group @end smallexample @need 2000 @node Recursive triangle function @subsection Recursion in Place of a Counter @findex triangle-recursively @need 1200 The @code{triangle} function described in a previous section can also be written recursively. It looks like this: @smallexample @group (defun triangle-recursively (number) "Return the sum of the numbers 1 through NUMBER inclusive. Uses recursion." (if (= number 1) ; @r{do-again-test} 1 ; @r{then-part} (+ number ; @r{else-part} (triangle-recursively ; @r{recursive call} (1- number))))) ; @r{next-step-expression} (triangle-recursively 7) @end group @end smallexample @noindent You can install this function by evaluating it and then try it by evaluating @code{(triangle-recursively 7)}. (Remember to put your cursor immediately after the last parenthesis of the function definition, before the comment.) The function evaluates to 28. To understand how this function works, let's consider what happens in the various cases when the function is passed 1, 2, 3, or 4 as the value of its argument. @menu * Recursive Example arg of 1 or 2:: * Recursive Example arg of 3 or 4:: @end menu @ifnottex @node Recursive Example arg of 1 or 2 @unnumberedsubsubsec An argument of 1 or 2 @end ifnottex First, what happens if the value of the argument is 1? The function has an @code{if} expression after the documentation string. It tests whether the value of @code{number} is equal to 1; if so, Emacs evaluates the then-part of the @code{if} expression, which returns the number 1 as the value of the function. (A triangle with one row has one pebble in it.) Suppose, however, that the value of the argument is 2. In this case, Emacs evaluates the else-part of the @code{if} expression. @need 1200 The else-part consists of an addition, the recursive call to @code{triangle-recursively} and a decrementing action; and it looks like this: @smallexample (+ number (triangle-recursively (1- number))) @end smallexample When Emacs evaluates this expression, the innermost expression is evaluated first; then the other parts in sequence. Here are the steps in detail: @table @i @item Step 1 @w{ } Evaluate the innermost expression. The innermost expression is @code{(1- number)} so Emacs decrements the value of @code{number} from 2 to 1. @item Step 2 @w{ } Evaluate the @code{triangle-recursively} function. The Lisp interpreter creates an individual instance of @code{triangle-recursively}. It does not matter that this function is contained within itself. Emacs passes the result Step 1 as the argument used by this instance of the @code{triangle-recursively} function In this case, Emacs evaluates @code{triangle-recursively} with an argument of 1. This means that this evaluation of @code{triangle-recursively} returns 1. @item Step 3 @w{ } Evaluate the value of @code{number}. The variable @code{number} is the second element of the list that starts with @code{+}; its value is 2. @item Step 4 @w{ } Evaluate the @code{+} expression. The @code{+} expression receives two arguments, the first from the evaluation of @code{number} (Step 3) and the second from the evaluation of @code{triangle-recursively} (Step 2). The result of the addition is the sum of 2 plus 1, and the number 3 is returned, which is correct. A triangle with two rows has three pebbles in it. @end table @node Recursive Example arg of 3 or 4 @unnumberedsubsubsec An argument of 3 or 4 Suppose that @code{triangle-recursively} is called with an argument of 3. @table @i @item Step 1 @w{ } Evaluate the do-again-test. The @code{if} expression is evaluated first. This is the do-again test and returns false, so the else-part of the @code{if} expression is evaluated. (Note that in this example, the do-again-test causes the function to call itself when it tests false, not when it tests true.) @item Step 2 @w{ } Evaluate the innermost expression of the else-part. The innermost expression of the else-part is evaluated, which decrements 3 to 2. This is the next-step-expression. @item Step 3 @w{ } Evaluate the @code{triangle-recursively} function. The number 2 is passed to the @code{triangle-recursively} function. We already know what happens when Emacs evaluates @code{triangle-recursively} with an argument of 2. After going through the sequence of actions described earlier, it returns a value of 3. So that is what will happen here. @item Step 4 @w{ } Evaluate the addition. 3 will be passed as an argument to the addition and will be added to the number with which the function was called, which is 3. @end table @noindent The value returned by the function as a whole will be 6. Now that we know what will happen when @code{triangle-recursively} is called with an argument of 3, it is evident what will happen if it is called with an argument of 4: @quotation @need 800 In the recursive call, the evaluation of @smallexample (triangle-recursively (1- 4)) @end smallexample @need 800 @noindent will return the value of evaluating @smallexample (triangle-recursively 3) @end smallexample @noindent which is 6 and this value will be added to 4 by the addition in the third line. @end quotation @noindent The value returned by the function as a whole will be 10. Each time @code{triangle-recursively} is evaluated, it evaluates a version of itself---a different instance of itself---with a smaller argument, until the argument is small enough so that it does not evaluate itself. Note that this particular design for a recursive function requires that operations be deferred. Before @code{(triangle-recursively 7)} can calculate its answer, it must call @code{(triangle-recursively 6)}; and before @code{(triangle-recursively 6)} can calculate its answer, it must call @code{(triangle-recursively 5)}; and so on. That is to say, the calculation that @code{(triangle-recursively 7)} makes must be deferred until @code{(triangle-recursively 6)} makes its calculation; and @code{(triangle-recursively 6)} must defer until @code{(triangle-recursively 5)} completes; and so on. If each of these instances of @code{triangle-recursively} are thought of as different robots, the first robot must wait for the second to complete its job, which must wait until the third completes, and so on. There is a way around this kind of waiting, which we will discuss in @ref{No Deferment, , Recursion without Deferments}. @node Recursion with cond @subsection Recursion Example Using @code{cond} @findex cond The version of @code{triangle-recursively} described earlier is written with the @code{if} special form. It can also be written using another special form called @code{cond}. The name of the special form @code{cond} is an abbreviation of the word @samp{conditional}. Although the @code{cond} special form is not used as often in the Emacs Lisp sources as @code{if}, it is used often enough to justify explaining it. @need 800 The template for a @code{cond} expression looks like this: @smallexample @group (cond @var{body}@dots{}) @end group @end smallexample @noindent where the @var{body} is a series of lists. @need 800 Written out more fully, the template looks like this: @smallexample @group (cond (@var{first-true-or-false-test} @var{first-consequent}) (@var{second-true-or-false-test} @var{second-consequent}) (@var{third-true-or-false-test} @var{third-consequent}) @dots{}) @end group @end smallexample When the Lisp interpreter evaluates the @code{cond} expression, it evaluates the first element (the @sc{car} or true-or-false-test) of the first expression in a series of expressions within the body of the @code{cond}. If the true-or-false-test returns @code{nil} the rest of that expression, the consequent, is skipped and the true-or-false-test of the next expression is evaluated. When an expression is found whose true-or-false-test returns a value that is not @code{nil}, the consequent of that expression is evaluated. The consequent can be one or more expressions. If the consequent consists of more than one expression, the expressions are evaluated in sequence and the value of the last one is returned. If the expression does not have a consequent, the value of the true-or-false-test is returned. If none of the true-or-false-tests test true, the @code{cond} expression returns @code{nil}. @need 1250 Written using @code{cond}, the @code{triangle} function looks like this: @smallexample @group (defun triangle-using-cond (number) (cond ((<= number 0) 0) ((= number 1) 1) ((> number 1) (+ number (triangle-using-cond (1- number)))))) @end group @end smallexample @noindent In this example, the @code{cond} returns 0 if the number is less than or equal to 0, it returns 1 if the number is 1 and it evaluates @code{(+ number (triangle-using-cond (1- number)))} if the number is greater than 1. @node Recursive Patterns @subsection Recursive Patterns @cindex Recursive Patterns Here are three common recursive patterns. Each involves a list. Recursion does not need to involve lists, but Lisp is designed for lists and this provides a sense of its primal capabilities. @menu * Every:: * Accumulate:: * Keep:: @end menu @node Every @unnumberedsubsubsec Recursive Pattern: @emph{every} @cindex Every, type of recursive pattern @cindex Recursive pattern - every In the @code{every} recursive pattern, an action is performed on every element of a list. @need 1500 The basic pattern is: @itemize @bullet @item If a list be empty, return @code{nil}. @item Else, act on the beginning of the list (the @sc{car} of the list) @itemize @minus @item through a recursive call by the function on the rest (the @sc{cdr}) of the list, @item and, optionally, combine the acted-on element, using @code{cons}, with the results of acting on the rest. @end itemize @end itemize @need 1500 Here is an example: @smallexample @group (defun square-each (numbers-list) "Square each of a NUMBERS LIST, recursively." (if (not numbers-list) ; do-again-test nil (cons (* (car numbers-list) (car numbers-list)) (square-each (cdr numbers-list))))) ; next-step-expression @end group @group (square-each '(1 2 3)) @result{} (1 4 9) @end group @end smallexample @need 1200 @noindent If @code{numbers-list} is empty, do nothing. But if it has content, construct a list combining the square of the first number in the list with the result of the recursive call. (The example follows the pattern exactly: @code{nil} is returned if the numbers' list is empty. In practice, you would write the conditional so it carries out the action when the numbers' list is not empty.) The @code{print-elements-recursively} function (@pxref{Recursion with list, , Recursion with a List}) is another example of an @code{every} pattern, except in this case, rather than bring the results together using @code{cons}, we print each element of output. @need 1250 The @code{print-elements-recursively} function looks like this: @smallexample @group (setq animals '(gazelle giraffe lion tiger)) @end group @group (defun print-elements-recursively (list) "Print each element of LIST on a line of its own. Uses recursion." (when list ; @r{do-again-test} (print (car list)) ; @r{body} (print-elements-recursively ; @r{recursive call} (cdr list)))) ; @r{next-step-expression} (print-elements-recursively animals) @end group @end smallexample @need 1500 The pattern for @code{print-elements-recursively} is: @itemize @bullet @item When the list is empty, do nothing. @item But when the list has at least one element, @itemize @minus @item act on the beginning of the list (the @sc{car} of the list), @item and make a recursive call on the rest (the @sc{cdr}) of the list. @end itemize @end itemize @node Accumulate @unnumberedsubsubsec Recursive Pattern: @emph{accumulate} @cindex Accumulate, type of recursive pattern @cindex Recursive pattern - accumulate Another recursive pattern is called the @code{accumulate} pattern. In the @code{accumulate} recursive pattern, an action is performed on every element of a list and the result of that action is accumulated with the results of performing the action on the other elements. This is very like the @code{every} pattern using @code{cons}, except that @code{cons} is not used, but some other combiner. @need 1500 The pattern is: @itemize @bullet @item If a list be empty, return zero or some other constant. @item Else, act on the beginning of the list (the @sc{car} of the list), @itemize @minus @item and combine that acted-on element, using @code{+} or some other combining function, with @item a recursive call by the function on the rest (the @sc{cdr}) of the list. @end itemize @end itemize @need 1500 Here is an example: @smallexample @group (defun add-elements (numbers-list) "Add the elements of NUMBERS-LIST together." (if (not numbers-list) 0 (+ (car numbers-list) (add-elements (cdr numbers-list))))) @end group @group (add-elements '(1 2 3 4)) @result{} 10 @end group @end smallexample @xref{Files List, , Making a List of Files}, for an example of the accumulate pattern. @node Keep @unnumberedsubsubsec Recursive Pattern: @emph{keep} @cindex Keep, type of recursive pattern @cindex Recursive pattern - keep A third recursive pattern is called the @code{keep} pattern. In the @code{keep} recursive pattern, each element of a list is tested; the element is acted on and the results are kept only if the element meets a criterion. Again, this is very like the @code{every} pattern, except the element is skipped unless it meets a criterion. @need 1500 The pattern has three parts: @itemize @bullet @item If a list be empty, return @code{nil}. @item Else, if the beginning of the list (the @sc{car} of the list) passes a test @itemize @minus @item act on that element and combine it, using @code{cons} with @item a recursive call by the function on the rest (the @sc{cdr}) of the list. @end itemize @item Otherwise, if the beginning of the list (the @sc{car} of the list) fails the test @itemize @minus @item skip on that element, @item and, recursively call the function on the rest (the @sc{cdr}) of the list. @end itemize @end itemize @need 1500 Here is an example that uses @code{cond}: @smallexample @group (defun keep-three-letter-words (word-list) "Keep three letter words in WORD-LIST." (cond ;; First do-again-test: stop-condition ((not word-list) nil) ;; Second do-again-test: when to act ((eq 3 (length (symbol-name (car word-list)))) ;; combine acted-on element with recursive call on shorter list (cons (car word-list) (keep-three-letter-words (cdr word-list)))) ;; Third do-again-test: when to skip element; ;; recursively call shorter list with next-step expression (t (keep-three-letter-words (cdr word-list))))) @end group @group (keep-three-letter-words '(one two three four five six)) @result{} (one two six) @end group @end smallexample It goes without saying that you need not use @code{nil} as the test for when to stop; and you can, of course, combine these patterns. @node No Deferment @subsection Recursion without Deferments @cindex Deferment in recursion @cindex Recursion without Deferments Let's consider again what happens with the @code{triangle-recursively} function. We will find that the intermediate calculations are deferred until all can be done. @need 800 Here is the function definition: @smallexample @group (defun triangle-recursively (number) "Return the sum of the numbers 1 through NUMBER inclusive. Uses recursion." (if (= number 1) ; @r{do-again-test} 1 ; @r{then-part} (+ number ; @r{else-part} (triangle-recursively ; @r{recursive call} (1- number))))) ; @r{next-step-expression} @end group @end smallexample What happens when we call this function with an argument of 7? The first instance of the @code{triangle-recursively} function adds the number 7 to the value returned by a second instance of @code{triangle-recursively}, an instance that has been passed an argument of 6. That is to say, the first calculation is: @smallexample (+ 7 (triangle-recursively 6)) @end smallexample @noindent The first instance of @code{triangle-recursively}---you may want to think of it as a little robot---cannot complete its job. It must hand off the calculation for @code{(triangle-recursively 6)} to a second instance of the program, to a second robot. This second individual is completely different from the first one; it is, in the jargon, a ``different instantiation''. Or, put another way, it is a different robot. It is the same model as the first; it calculates triangle numbers recursively; but it has a different serial number. And what does @code{(triangle-recursively 6)} return? It returns the number 6 added to the value returned by evaluating @code{triangle-recursively} with an argument of 5. Using the robot metaphor, it asks yet another robot to help it. @need 800 Now the total is: @smallexample (+ 7 6 (triangle-recursively 5)) @end smallexample @need 800 And what happens next? @smallexample (+ 7 6 5 (triangle-recursively 4)) @end smallexample Each time @code{triangle-recursively} is called, except for the last time, it creates another instance of the program---another robot---and asks it to make a calculation. @need 800 Eventually, the full addition is set up and performed: @smallexample (+ 7 6 5 4 3 2 1) @end smallexample This design for the function defers the calculation of the first step until the second can be done, and defers that until the third can be done, and so on. Each deferment means the computer must remember what is being waited on. This is not a problem when there are only a few steps, as in this example. But it can be a problem when there are more steps. @node No deferment solution @subsection No Deferment Solution @cindex No deferment solution @cindex Solution without deferment The solution to the problem of deferred operations is to write in a manner that does not defer operations@footnote{The phrase @dfn{tail recursive} is used to describe such a process, one that uses constant space.}. This requires writing to a different pattern, often one that involves writing two function definitions, an initialization function and a helper function. The initialization function sets up the job; the helper function does the work. @need 1200 Here are the two function definitions for adding up numbers. They are so simple, I find them hard to understand. @smallexample @group (defun triangle-initialization (number) "Return the sum of the numbers 1 through NUMBER inclusive. This is the initialization component of a two function duo that uses recursion." (triangle-recursive-helper 0 0 number)) @end group @end smallexample @smallexample @group (defun triangle-recursive-helper (sum counter number) "Return SUM, using COUNTER, through NUMBER inclusive. This is the helper component of a two function duo that uses recursion." (if (> counter number) sum (triangle-recursive-helper (+ sum counter) ; @r{sum} (1+ counter) ; @r{counter} number))) ; @r{number} @end group @end smallexample @need 1250 Install both function definitions by evaluating them, then call @code{triangle-initialization} with 2 rows: @smallexample @group (triangle-initialization 2) @result{} 3 @end group @end smallexample The initialization function calls the first instance of the helper function with three arguments: zero, zero, and a number which is the number of rows in the triangle. The first two arguments passed to the helper function are initialization values. These values are changed when @code{triangle-recursive-helper} invokes new instances.@footnote{The jargon is mildly confusing: @code{triangle-recursive-helper} uses a process that is iterative in a procedure that is recursive. The process is called iterative because the computer need only record the three values, @code{sum}, @code{counter}, and @code{number}; the procedure is recursive because the function calls itself. On the other hand, both the process and the procedure used by @code{triangle-recursively} are called recursive. The word ``recursive'' has different meanings in the two contexts.} Let's see what happens when we have a triangle that has one row. (This triangle will have one pebble in it!) @need 1200 @code{triangle-initialization} will call its helper with the arguments @w{@code{0 0 1}}. That function will run the conditional test whether @code{(> counter number)}: @smallexample (> 0 1) @end smallexample @need 1200 @noindent and find that the result is false, so it will invoke the else-part of the @code{if} clause: @smallexample @group (triangle-recursive-helper (+ sum counter) ; @r{sum plus counter} @result{} @r{sum} (1+ counter) ; @r{increment counter} @result{} @r{counter} number) ; @r{number stays the same} @end group @end smallexample @need 800 @noindent which will first compute: @smallexample @group (triangle-recursive-helper (+ 0 0) ; @r{sum} (1+ 0) ; @r{counter} 1) ; @r{number} @exdent which is: (triangle-recursive-helper 0 1 1) @end group @end smallexample Again, @code{(> counter number)} will be false, so again, the Lisp interpreter will evaluate @code{triangle-recursive-helper}, creating a new instance with new arguments. @need 800 This new instance will be; @smallexample @group (triangle-recursive-helper (+ sum counter) ; @r{sum plus counter} @result{} @r{sum} (1+ counter) ; @r{increment counter} @result{} @r{counter} number) ; @r{number stays the same} @exdent which is: (triangle-recursive-helper 1 2 1) @end group @end smallexample In this case, the @code{(> counter number)} test will be true! So the instance will return the value of the sum, which will be 1, as expected. Now, let's pass @code{triangle-initialization} an argument of 2, to find out how many pebbles there are in a triangle with two rows. That function calls @code{(triangle-recursive-helper 0 0 2)}. @need 800 In stages, the instances called will be: @smallexample @group @r{sum counter number} (triangle-recursive-helper 0 1 2) (triangle-recursive-helper 1 2 2) (triangle-recursive-helper 3 3 2) @end group @end smallexample When the last instance is called, the @code{(> counter number)} test will be true, so the instance will return the value of @code{sum}, which will be 3. This kind of pattern helps when you are writing functions that can use many resources in a computer. @need 1500 @node Looping exercise @section Looping Exercise @itemize @bullet @item Write a function similar to @code{triangle} in which each row has a value which is the square of the row number. Use a @code{while} loop. @item Write a function similar to @code{triangle} that multiplies instead of adds the values. @item Rewrite these two functions recursively. Rewrite these functions using @code{cond}. @c comma in printed title causes problem in Info cross reference @item Write a function for Texinfo mode that creates an index entry at the beginning of a paragraph for every @samp{@@dfn} within the paragraph. (In a Texinfo file, @samp{@@dfn} marks a definition. This book is written in Texinfo.) Many of the functions you will need are described in two of the previous chapters, @ref{Cutting & Storing Text, , Cutting and Storing Text}, and @ref{Yanking, , Yanking Text Back}. If you use @code{forward-paragraph} to put the index entry at the beginning of the paragraph, you will have to use @w{@kbd{C-h f}} (@code{describe-function}) to find out how to make the command go backwards. For more information, see @ifinfo @ref{Indicating, , Indicating Definitions, texinfo}. @end ifinfo @ifhtml @ref{Indicating, , Indicating, texinfo, Texinfo Manual}, which goes to a Texinfo manual in the current directory. Or, if you are on the Internet, see @uref{https://www.gnu.org/software/texinfo/manual/texinfo/} @end ifhtml @iftex ``Indicating Definitions, Commands, etc.''@: in @cite{Texinfo, The GNU Documentation Format}. @end iftex @end itemize @node Regexp Search @chapter Regular Expression Searches @cindex Searches, illustrating @cindex Regular expression searches @cindex Patterns, searching for @cindex Motion by sentence and paragraph @cindex Sentences, movement by @cindex Paragraphs, movement by Regular expression searches are used extensively in GNU Emacs. The two functions, @code{forward-sentence} and @code{forward-paragraph}, illustrate these searches well. They use regular expressions to find where to move point. The phrase ``regular expression'' is often written as ``regexp''. Regular expression searches are described in @ref{Regexp Search, , Regular Expression Search, emacs, The GNU Emacs Manual}, as well as in @ref{Regular Expressions, , , elisp, The GNU Emacs Lisp Reference Manual}. In writing this chapter, I am presuming that you have at least a mild acquaintance with them. The major point to remember is that regular expressions permit you to search for patterns as well as for literal strings of characters. For example, the code in @code{forward-sentence} searches for the pattern of possible characters that could mark the end of a sentence, and moves point to that spot. Before looking at the code for the @code{forward-sentence} function, it is worth considering what the pattern that marks the end of a sentence must be. The pattern is discussed in the next section; following that is a description of the regular expression search function, @code{re-search-forward}. The @code{forward-sentence} function is described in the section following. Finally, the @code{forward-paragraph} function is described in the last section of this chapter. @code{forward-paragraph} is a complex function that introduces several new features. @menu * sentence-end:: The regular expression for @code{sentence-end}. * re-search-forward:: Very similar to @code{search-forward}. * forward-sentence:: A straightforward example of regexp search. * forward-paragraph:: A somewhat complex example. * Regexp Review:: * re-search Exercises:: @end menu @node sentence-end @section The Regular Expression for @code{sentence-end} @findex sentence-end The symbol @code{sentence-end} is bound to the pattern that marks the end of a sentence. What should this regular expression be? Clearly, a sentence may be ended by a period, a question mark, or an exclamation mark. Indeed, in English, only clauses that end with one of those three characters should be considered the end of a sentence. This means that the pattern should include the character set: @smallexample [.?!] @end smallexample However, we do not want @code{forward-sentence} merely to jump to a period, a question mark, or an exclamation mark, because such a character might be used in the middle of a sentence. A period, for example, is used after abbreviations. So other information is needed. According to convention, you type two spaces after every sentence, but only one space after a period, a question mark, or an exclamation mark in the body of a sentence. So a period, a question mark, or an exclamation mark followed by two spaces is a good indicator of an end of sentence. However, in a file, the two spaces may instead be a tab or the end of a line. This means that the regular expression should include these three items as alternatives. @need 800 This group of alternatives will look like this: @smallexample @group \\($\\| \\| \\) ^ ^^ TAB SPC @end group @end smallexample @noindent Here, @samp{$} indicates the end of the line, and I have pointed out where the tab and two spaces are inserted in the expression. Both are inserted by putting the actual characters into the expression. Two backslashes, @samp{\\}, are required before the parentheses and vertical bars: the first backslash quotes the following backslash in Emacs; and the second indicates that the following character, the parenthesis or the vertical bar, is special. @need 1000 Also, a sentence may be followed by one or more carriage returns, like this: @smallexample @group [ ]* @end group @end smallexample @noindent Like tabs and spaces, a carriage return is inserted into a regular expression by inserting it literally. The asterisk indicates that the @key{RET} is repeated zero or more times. But a sentence end does not consist only of a period, a question mark or an exclamation mark followed by appropriate space: a closing quotation mark or a closing brace of some kind may precede the space. Indeed more than one such mark or brace may precede the space. These require a expression that looks like this: @smallexample []\"')@}]* @end smallexample In this expression, the first @samp{]} is the first character in the expression; the second character is @samp{"}, which is preceded by a @samp{\} to tell Emacs the @samp{"} is @emph{not} special. The last three characters are @samp{'}, @samp{)}, and @samp{@}}. All this suggests what the regular expression pattern for matching the end of a sentence should be; and, indeed, if we evaluate @code{sentence-end} we find that it returns the following value: @smallexample @group sentence-end @result{} "[.?!][]\"')@}]*\\($\\| \\| \\)[ ]*" @end group @end smallexample @noindent (Well, not in GNU Emacs 22; that is because of an effort to make the process simpler and to handle more glyphs and languages. When the value of @code{sentence-end} is @code{nil}, then use the value defined by the function @code{sentence-end}. (Here is a use of the difference between a value and a function in Emacs Lisp.) The function returns a value constructed from the variables @code{sentence-end-base}, @code{sentence-end-double-space}, @code{sentence-end-without-period}, and @code{sentence-end-without-space}. The critical variable is @code{sentence-end-base}; its global value is similar to the one described above but it also contains two additional quotation marks. These have differing degrees of curliness. The @code{sentence-end-without-period} variable, when true, tells Emacs that a sentence may end without a period, such as text in Thai.) @ignore @noindent (Note that here the @key{TAB}, two spaces, and @key{RET} are shown literally in the pattern.) This regular expression can be deciphered as follows: @table @code @item [.?!] The first part of the pattern is the three characters, a period, a question mark and an exclamation mark, within square brackets. The pattern must begin with one or other of these characters. @item []\"')@}]* The second part of the pattern is the group of closing braces and quotation marks, which can appear zero or more times. These may follow the period, question mark or exclamation mark. In a regular expression, the backslash, @samp{\}, followed by the double quotation mark, @samp{"}, indicates the class of string-quote characters. Usually, the double quotation mark is the only character in this class. The asterisk, @samp{*}, indicates that the items in the previous group (the group surrounded by square brackets, @samp{[]}) may be repeated zero or more times. @item \\($\\| \\| \\) The third part of the pattern is one or other of: either the end of a line, or two blank spaces, or a tab. The double back-slashes are used to prevent Emacs from reading the parentheses and vertical bars as part of the search pattern; the parentheses are used to mark the group and the vertical bars are used to indicated that the patterns to either side of them are alternatives. The dollar sign is used to indicate the end of a line and both the two spaces and the tab are each inserted as is to indicate what they are. @item [@key{RET}]* Finally, the last part of the pattern indicates that the end of the line or the whitespace following the period, question mark or exclamation mark may, but need not, be followed by one or more carriage returns. In the pattern, the carriage return is inserted as an actual carriage return between square brackets but here it is shown as @key{RET}. @end table @end ignore @node re-search-forward @section The @code{re-search-forward} Function @findex re-search-forward The @code{re-search-forward} function is very like the @code{search-forward} function. (@xref{search-forward, , The @code{search-forward} Function}.) @code{re-search-forward} searches for a regular expression. If the search is successful, it leaves point immediately after the last character in the target. If the search is backwards, it leaves point just before the first character in the target. You may tell @code{re-search-forward} to return @code{t} for true. (Moving point is therefore a side effect.) Like @code{search-forward}, the @code{re-search-forward} function takes four arguments: @enumerate @item The first argument is the regular expression that the function searches for. The regular expression will be a string between quotation marks. @item The optional second argument limits how far the function will search; it is a bound, which is specified as a position in the buffer. @item The optional third argument specifies how the function responds to failure: @code{nil} as the third argument causes the function to signal an error (and print a message) when the search fails; any other value causes it to return @code{nil} if the search fails and @code{t} if the search succeeds. @item The optional fourth argument is the repeat count. A negative repeat count causes @code{re-search-forward} to search backwards. @end enumerate @need 800 The template for @code{re-search-forward} looks like this: @smallexample @group (re-search-forward "@var{regular-expression}" @var{limit-of-search} @var{what-to-do-if-search-fails} @var{repeat-count}) @end group @end smallexample The second, third, and fourth arguments are optional. However, if you want to pass a value to either or both of the last two arguments, you must also pass a value to all the preceding arguments. Otherwise, the Lisp interpreter will mistake which argument you are passing the value to. @need 1200 In the @code{forward-sentence} function, the regular expression will be the value of the variable @code{sentence-end}. In simple form, that is: @smallexample @group "[.?!][]\"')@}]*\\($\\| \\| \\)[ ]*" @end group @end smallexample @noindent The limit of the search will be the end of the paragraph (since a sentence cannot go beyond a paragraph). If the search fails, the function will return @code{nil}; and the repeat count will be provided by the argument to the @code{forward-sentence} function. @node forward-sentence @section @code{forward-sentence} @findex forward-sentence The command to move the cursor forward a sentence is a straightforward illustration of how to use regular expression searches in Emacs Lisp. Indeed, the function looks longer and more complicated than it is; this is because the function is designed to go backwards as well as forwards; and, optionally, over more than one sentence. The function is usually bound to the key command @kbd{M-e}. @menu * Complete forward-sentence:: * fwd-sentence while loops:: Two @code{while} loops. * fwd-sentence re-search:: A regular expression search. @end menu @ifnottex @node Complete forward-sentence @unnumberedsubsec Complete @code{forward-sentence} function definition @end ifnottex @need 1250 Here is the code for @code{forward-sentence}: @c in GNU Emacs 22 @smallexample @group (defun forward-sentence (&optional arg) "Move forward to next end of sentence. With argument, repeat. With negative argument, move backward repeatedly to start of sentence. The variable `sentence-end' is a regular expression that matches ends of sentences. Also, every paragraph boundary terminates sentences as well." @end group @group (interactive "p") (or arg (setq arg 1)) (let ((opoint (point)) (sentence-end (sentence-end))) (while (< arg 0) (let ((pos (point)) (par-beg (save-excursion (start-of-paragraph-text) (point)))) (if (and (re-search-backward sentence-end par-beg t) (or (< (match-end 0) pos) (re-search-backward sentence-end par-beg t))) (goto-char (match-end 0)) (goto-char par-beg))) (setq arg (1+ arg))) @end group @group (while (> arg 0) (let ((par-end (save-excursion (end-of-paragraph-text) (point)))) (if (re-search-forward sentence-end par-end t) (skip-chars-backward " \t\n") (goto-char par-end))) (setq arg (1- arg))) (constrain-to-field nil opoint t))) @end group @end smallexample @ignore GNU Emacs 21 @smallexample @group (defun forward-sentence (&optional arg) "Move forward to next sentence-end. With argument, repeat. With negative argument, move backward repeatedly to sentence-beginning. Sentence ends are identified by the value of sentence-end treated as a regular expression. Also, every paragraph boundary terminates sentences as well." @end group @group (interactive "p") (or arg (setq arg 1)) (while (< arg 0) (let ((par-beg (save-excursion (start-of-paragraph-text) (point)))) (if (re-search-backward (concat sentence-end "[^ \t\n]") par-beg t) (goto-char (1- (match-end 0))) (goto-char par-beg))) (setq arg (1+ arg))) (while (> arg 0) (let ((par-end (save-excursion (end-of-paragraph-text) (point)))) (if (re-search-forward sentence-end par-end t) (skip-chars-backward " \t\n") (goto-char par-end))) (setq arg (1- arg)))) @end group @end smallexample @end ignore The function looks long at first sight and it is best to look at its skeleton first, and then its muscle. The way to see the skeleton is to look at the expressions that start in the left-most columns: @smallexample @group (defun forward-sentence (&optional arg) "@var{documentation}@dots{}" (interactive "p") (or arg (setq arg 1)) (let ((opoint (point)) (sentence-end (sentence-end))) (while (< arg 0) (let ((pos (point)) (par-beg (save-excursion (start-of-paragraph-text) (point)))) @var{rest-of-body-of-while-loop-when-going-backwards} (while (> arg 0) (let ((par-end (save-excursion (end-of-paragraph-text) (point)))) @var{rest-of-body-of-while-loop-when-going-forwards} @var{handle-forms-and-equivalent} @end group @end smallexample This looks much simpler! The function definition consists of documentation, an @code{interactive} expression, an @code{or} expression, a @code{let} expression, and @code{while} loops. Let's look at each of these parts in turn. We note that the documentation is thorough and understandable. The function has an @code{interactive "p"} declaration. This means that the processed prefix argument, if any, is passed to the function as its argument. (This will be a number.) If the function is not passed an argument (it is optional) then the argument @code{arg} will be bound to 1. When @code{forward-sentence} is called non-interactively without an argument, @code{arg} is bound to @code{nil}. The @code{or} expression handles this. What it does is either leave the value of @code{arg} as it is, but only if @code{arg} is bound to a value; or it sets the value of @code{arg} to 1, in the case when @code{arg} is bound to @code{nil}. Next is a @code{let}. That specifies the values of two local variables, @code{opoint} and @code{sentence-end}. The local value of point, from before the search, is used in the @code{constrain-to-field} function which handles forms and equivalents. The @code{sentence-end} variable is set by the @code{sentence-end} function. @node fwd-sentence while loops @unnumberedsubsec The @code{while} loops Two @code{while} loops follow. The first @code{while} has a true-or-false-test that tests true if the prefix argument for @code{forward-sentence} is a negative number. This is for going backwards. The body of this loop is similar to the body of the second @code{while} clause, but it is not exactly the same. We will skip this @code{while} loop and concentrate on the second @code{while} loop. @need 1500 The second @code{while} loop is for moving point forward. Its skeleton looks like this: @smallexample @group (while (> arg 0) ; @r{true-or-false-test} (let @var{varlist} (if (@var{true-or-false-test}) @var{then-part} @var{else-part} (setq arg (1- arg)))) ; @code{while} @r{loop decrementer} @end group @end smallexample The @code{while} loop is of the decrementing kind. (@xref{Decrementing Loop, , A Loop with a Decrementing Counter}.) It has a true-or-false-test that tests true so long as the counter (in this case, the variable @code{arg}) is greater than zero; and it has a decrementer that subtracts 1 from the value of the counter every time the loop repeats. If no prefix argument is given to @code{forward-sentence}, which is the most common way the command is used, this @code{while} loop will run once, since the value of @code{arg} will be 1. The body of the @code{while} loop consists of a @code{let} expression, which creates and binds a local variable, and has, as its body, an @code{if} expression. @need 1250 The body of the @code{while} loop looks like this: @smallexample @group (let ((par-end (save-excursion (end-of-paragraph-text) (point)))) (if (re-search-forward sentence-end par-end t) (skip-chars-backward " \t\n") (goto-char par-end))) @end group @end smallexample The @code{let} expression creates and binds the local variable @code{par-end}. As we shall see, this local variable is designed to provide a bound or limit to the regular expression search. If the search fails to find a proper sentence ending in the paragraph, it will stop on reaching the end of the paragraph. But first, let us examine how @code{par-end} is bound to the value of the end of the paragraph. What happens is that the @code{let} sets the value of @code{par-end} to the value returned when the Lisp interpreter evaluates the expression @smallexample @group (save-excursion (end-of-paragraph-text) (point)) @end group @end smallexample @noindent In this expression, @code{(end-of-paragraph-text)} moves point to the end of the paragraph, @code{(point)} returns the value of point, and then @code{save-excursion} restores point to its original position. Thus, the @code{let} binds @code{par-end} to the value returned by the @code{save-excursion} expression, which is the position of the end of the paragraph. (The @code{end-of-paragraph-text} function uses @code{forward-paragraph}, which we will discuss shortly.) @need 1200 Emacs next evaluates the body of the @code{let}, which is an @code{if} expression that looks like this: @smallexample @group (if (re-search-forward sentence-end par-end t) ; @r{if-part} (skip-chars-backward " \t\n") ; @r{then-part} (goto-char par-end))) ; @r{else-part} @end group @end smallexample The @code{if} tests whether its first argument is true and if so, evaluates its then-part; otherwise, the Emacs Lisp interpreter evaluates the else-part. The true-or-false-test of the @code{if} expression is the regular expression search. It may seem odd to have what looks like the real work of the @code{forward-sentence} function buried here, but this is a common way this kind of operation is carried out in Lisp. @node fwd-sentence re-search @unnumberedsubsec The regular expression search The @code{re-search-forward} function searches for the end of the sentence, that is, for the pattern defined by the @code{sentence-end} regular expression. If the pattern is found---if the end of the sentence is found---then the @code{re-search-forward} function does two things: @enumerate @item The @code{re-search-forward} function carries out a side effect, which is to move point to the end of the occurrence found. @item The @code{re-search-forward} function returns a value of true. This is the value received by the @code{if}, and means that the search was successful. @end enumerate @noindent The side effect, the movement of point, is completed before the @code{if} function is handed the value returned by the successful conclusion of the search. When the @code{if} function receives the value of true from a successful call to @code{re-search-forward}, the @code{if} evaluates the then-part, which is the expression @code{(skip-chars-backward " \t\n")}. This expression moves backwards over any blank spaces, tabs or carriage returns until a printed character is found and then leaves point after the character. Since point has already been moved to the end of the pattern that marks the end of the sentence, this action leaves point right after the closing printed character of the sentence, which is usually a period. On the other hand, if the @code{re-search-forward} function fails to find a pattern marking the end of the sentence, the function returns false. The false then causes the @code{if} to evaluate its third argument, which is @code{(goto-char par-end)}: it moves point to the end of the paragraph. (And if the text is in a form or equivalent, and point may not move fully, then the @code{constrain-to-field} function comes into play.) Regular expression searches are exceptionally useful and the pattern illustrated by @code{re-search-forward}, in which the search is the test of an @code{if} expression, is handy. You will see or write code incorporating this pattern often. @node forward-paragraph @section @code{forward-paragraph}: a Goldmine of Functions @findex forward-paragraph @ignore @c in GNU Emacs 22 (defun forward-paragraph (&optional arg) "Move forward to end of paragraph. With argument ARG, do it ARG times; a negative argument ARG = -N means move backward N paragraphs. A line which `paragraph-start' matches either separates paragraphs \(if `paragraph-separate' matches it also) or is the first line of a paragraph. A paragraph end is the beginning of a line which is not part of the paragraph to which the end of the previous line belongs, or the end of the buffer. Returns the count of paragraphs left to move." (interactive "p") (or arg (setq arg 1)) (let* ((opoint (point)) (fill-prefix-regexp (and fill-prefix (not (equal fill-prefix "")) (not paragraph-ignore-fill-prefix) (regexp-quote fill-prefix))) ;; Remove ^ from paragraph-start and paragraph-sep if they are there. ;; These regexps shouldn't be anchored, because we look for them ;; starting at the left-margin. This allows paragraph commands to ;; work normally with indented text. ;; This hack will not find problem cases like "whatever\\|^something". (parstart (if (and (not (equal "" paragraph-start)) (equal ?^ (aref paragraph-start 0))) (substring paragraph-start 1) paragraph-start)) (parsep (if (and (not (equal "" paragraph-separate)) (equal ?^ (aref paragraph-separate 0))) (substring paragraph-separate 1) paragraph-separate)) (parsep (if fill-prefix-regexp (concat parsep "\\|" fill-prefix-regexp "[ \t]*$") parsep)) ;; This is used for searching. (sp-parstart (concat "^[ \t]*\\(?:" parstart "\\|" parsep "\\)")) start found-start) (while (and (< arg 0) (not (bobp))) (if (and (not (looking-at parsep)) (re-search-backward "^\n" (max (1- (point)) (point-min)) t) (looking-at parsep)) (setq arg (1+ arg)) (setq start (point)) ;; Move back over paragraph-separating lines. (forward-char -1) (beginning-of-line) (while (and (not (bobp)) (progn (move-to-left-margin) (looking-at parsep))) (forward-line -1)) (if (bobp) nil (setq arg (1+ arg)) ;; Go to end of the previous (non-separating) line. (end-of-line) ;; Search back for line that starts or separates paragraphs. (if (if fill-prefix-regexp ;; There is a fill prefix; it overrides parstart. (let (multiple-lines) (while (and (progn (beginning-of-line) (not (bobp))) (progn (move-to-left-margin) (not (looking-at parsep))) (looking-at fill-prefix-regexp)) (unless (= (point) start) (setq multiple-lines t)) (forward-line -1)) (move-to-left-margin) ;; This deleted code caused a long hanging-indent line ;; not to be filled together with the following lines. ;; ;; Don't move back over a line before the paragraph ;; ;; which doesn't start with fill-prefix ;; ;; unless that is the only line we've moved over. ;; (and (not (looking-at fill-prefix-regexp)) ;; multiple-lines ;; (forward-line 1)) (not (bobp))) (while (and (re-search-backward sp-parstart nil 1) (setq found-start t) ;; Found a candidate, but need to check if it is a ;; REAL parstart. (progn (setq start (point)) (move-to-left-margin) (not (looking-at parsep))) (not (and (looking-at parstart) (or (not use-hard-newlines) (bobp) (get-text-property (1- start) 'hard))))) (setq found-start nil) (goto-char start)) found-start) ;; Found one. (progn ;; Move forward over paragraph separators. ;; We know this cannot reach the place we started ;; because we know we moved back over a non-separator. (while (and (not (eobp)) (progn (move-to-left-margin) (looking-at parsep))) (forward-line 1)) ;; If line before paragraph is just margin, back up to there. (end-of-line 0) (if (> (current-column) (current-left-margin)) (forward-char 1) (skip-chars-backward " \t") (if (not (bolp)) (forward-line 1)))) ;; No starter or separator line => use buffer beg. (goto-char (point-min)))))) (while (and (> arg 0) (not (eobp))) ;; Move forward over separator lines... (while (and (not (eobp)) (progn (move-to-left-margin) (not (eobp))) (looking-at parsep)) (forward-line 1)) (unless (eobp) (setq arg (1- arg))) ;; ... and one more line. (forward-line 1) (if fill-prefix-regexp ;; There is a fill prefix; it overrides parstart. (while (and (not (eobp)) (progn (move-to-left-margin) (not (eobp))) (not (looking-at parsep)) (looking-at fill-prefix-regexp)) (forward-line 1)) (while (and (re-search-forward sp-parstart nil 1) (progn (setq start (match-beginning 0)) (goto-char start) (not (eobp))) (progn (move-to-left-margin) (not (looking-at parsep))) (or (not (looking-at parstart)) (and use-hard-newlines (not (get-text-property (1- start) 'hard))))) (forward-char 1)) (if (< (point) (point-max)) (goto-char start)))) (constrain-to-field nil opoint t) ;; Return the number of steps that could not be done. arg)) @end ignore The @code{forward-paragraph} function moves point forward to the end of the paragraph. It is usually bound to @kbd{M-@}} and makes use of a number of functions that are important in themselves, including @code{let*}, @code{match-beginning}, and @code{looking-at}. The function definition for @code{forward-paragraph} is considerably longer than the function definition for @code{forward-sentence} because it works with a paragraph, each line of which may begin with a fill prefix. A fill prefix consists of a string of characters that are repeated at the beginning of each line. For example, in Lisp code, it is a convention to start each line of a paragraph-long comment with @samp{;;; }. In Text mode, four blank spaces make up another common fill prefix, creating an indented paragraph. (@xref{Fill Prefix, , , emacs, The GNU Emacs Manual}, for more information about fill prefixes.) The existence of a fill prefix means that in addition to being able to find the end of a paragraph whose lines begin on the left-most column, the @code{forward-paragraph} function must be able to find the end of a paragraph when all or many of the lines in the buffer begin with the fill prefix. Moreover, it is sometimes practical to ignore a fill prefix that exists, especially when blank lines separate paragraphs. This is an added complication. @menu * forward-paragraph in brief:: Key parts of the function definition. * fwd-para let:: The @code{let*} expression. * fwd-para while:: The forward motion @code{while} loop. @end menu @ifnottex @node forward-paragraph in brief @unnumberedsubsec Shortened @code{forward-paragraph} function definition @end ifnottex Rather than print all of the @code{forward-paragraph} function, we will only print parts of it. Read without preparation, the function can be daunting! @need 800 In outline, the function looks like this: @smallexample @group (defun forward-paragraph (&optional arg) "@var{documentation}@dots{}" (interactive "p") (or arg (setq arg 1)) (let* @var{varlist} (while (and (< arg 0) (not (bobp))) ; @r{backward-moving-code} @dots{} (while (and (> arg 0) (not (eobp))) ; @r{forward-moving-code} @dots{} @end group @end smallexample The first parts of the function are routine: the function's argument list consists of one optional argument. Documentation follows. The lower case @samp{p} in the @code{interactive} declaration means that the processed prefix argument, if any, is passed to the function. This will be a number, and is the repeat count of how many paragraphs point will move. The @code{or} expression in the next line handles the common case when no argument is passed to the function, which occurs if the function is called from other code rather than interactively. This case was described earlier. (@xref{forward-sentence, The @code{forward-sentence} function}.) Now we reach the end of the familiar part of this function. @node fwd-para let @unnumberedsubsec The @code{let*} expression The next line of the @code{forward-paragraph} function begins a @code{let*} expression. This is different from @code{let}. The symbol is @code{let*} not @code{let}. @findex let* The @code{let*} special form is like @code{let} except that Emacs sets each variable in sequence, one after another, and variables in the latter part of the varlist can make use of the values to which Emacs set variables in the earlier part of the varlist. @ignore ( refappend save-excursion, , code save-excursion in code append-to-buffer .) @end ignore (@ref{append save-excursion, , @code{save-excursion} in @code{append-to-buffer}}.) In the @code{let*} expression in this function, Emacs binds a total of seven variables: @code{opoint}, @code{fill-prefix-regexp}, @code{parstart}, @code{parsep}, @code{sp-parstart}, @code{start}, and @code{found-start}. The variable @code{parsep} appears twice, first, to remove instances of @samp{^}, and second, to handle fill prefixes. The variable @code{opoint} is just the value of @code{point}. As you can guess, it is used in a @code{constrain-to-field} expression, just as in @code{forward-sentence}. The variable @code{fill-prefix-regexp} is set to the value returned by evaluating the following list: @smallexample @group (and fill-prefix (not (equal fill-prefix "")) (not paragraph-ignore-fill-prefix) (regexp-quote fill-prefix)) @end group @end smallexample @noindent This is an expression whose first element is the @code{and} special form. As we learned earlier (@pxref{kill-new function, , The @code{kill-new} function}), the @code{and} special form evaluates each of its arguments until one of the arguments returns a value of @code{nil}, in which case the @code{and} expression returns @code{nil}; however, if none of the arguments returns a value of @code{nil}, the value resulting from evaluating the last argument is returned. (Since such a value is not @code{nil}, it is considered true in Lisp.) In other words, an @code{and} expression returns a true value only if all its arguments are true. @findex and In this case, the variable @code{fill-prefix-regexp} is bound to a non-@code{nil} value only if the following four expressions produce a true (i.e., a non-@code{nil}) value when they are evaluated; otherwise, @code{fill-prefix-regexp} is bound to @code{nil}. @table @code @item fill-prefix When this variable is evaluated, the value of the fill prefix, if any, is returned. If there is no fill prefix, this variable returns @code{nil}. @item (not (equal fill-prefix "") This expression checks whether an existing fill prefix is an empty string, that is, a string with no characters in it. An empty string is not a useful fill prefix. @item (not paragraph-ignore-fill-prefix) This expression returns @code{nil} if the variable @code{paragraph-ignore-fill-prefix} has been turned on by being set to a true value such as @code{t}. @item (regexp-quote fill-prefix) This is the last argument to the @code{and} special form. If all the arguments to the @code{and} are true, the value resulting from evaluating this expression will be returned by the @code{and} expression and bound to the variable @code{fill-prefix-regexp}, @end table @findex regexp-quote @noindent The result of evaluating this @code{and} expression successfully is that @code{fill-prefix-regexp} will be bound to the value of @code{fill-prefix} as modified by the @code{regexp-quote} function. What @code{regexp-quote} does is read a string and return a regular expression that will exactly match the string and match nothing else. This means that @code{fill-prefix-regexp} will be set to a value that will exactly match the fill prefix if the fill prefix exists. Otherwise, the variable will be set to @code{nil}. The next two local variables in the @code{let*} expression are designed to remove instances of @samp{^} from @code{parstart} and @code{parsep}, the local variables which indicate the paragraph start and the paragraph separator. The next expression sets @code{parsep} again. That is to handle fill prefixes. This is the setting that requires the definition call @code{let*} rather than @code{let}. The true-or-false-test for the @code{if} depends on whether the variable @code{fill-prefix-regexp} evaluates to @code{nil} or some other value. If @code{fill-prefix-regexp} does not have a value, Emacs evaluates the else-part of the @code{if} expression and binds @code{parsep} to its local value. (@code{parsep} is a regular expression that matches what separates paragraphs.) But if @code{fill-prefix-regexp} does have a value, Emacs evaluates the then-part of the @code{if} expression and binds @code{parsep} to a regular expression that includes the @code{fill-prefix-regexp} as part of the pattern. Specifically, @code{parsep} is set to the original value of the paragraph separate regular expression concatenated with an alternative expression that consists of the @code{fill-prefix-regexp} followed by optional whitespace to the end of the line. The whitespace is defined by @w{@code{"[ \t]*$"}}.) The @samp{\\|} defines this portion of the regexp as an alternative to @code{parsep}. According to a comment in the code, the next local variable, @code{sp-parstart}, is used for searching, and then the final two, @code{start} and @code{found-start}, are set to @code{nil}. Now we get into the body of the @code{let*}. The first part of the body of the @code{let*} deals with the case when the function is given a negative argument and is therefore moving backwards. We will skip this section. @node fwd-para while @unnumberedsubsec The forward motion @code{while} loop The second part of the body of the @code{let*} deals with forward motion. It is a @code{while} loop that repeats itself so long as the value of @code{arg} is greater than zero. In the most common use of the function, the value of the argument is 1, so the body of the @code{while} loop is evaluated exactly once, and the cursor moves forward one paragraph. @ignore (while (and (> arg 0) (not (eobp))) ;; Move forward over separator lines... (while (and (not (eobp)) (progn (move-to-left-margin) (not (eobp))) (looking-at parsep)) (forward-line 1)) (unless (eobp) (setq arg (1- arg))) ;; ... and one more line. (forward-line 1) (if fill-prefix-regexp ;; There is a fill prefix; it overrides parstart. (while (and (not (eobp)) (progn (move-to-left-margin) (not (eobp))) (not (looking-at parsep)) (looking-at fill-prefix-regexp)) (forward-line 1)) (while (and (re-search-forward sp-parstart nil 1) (progn (setq start (match-beginning 0)) (goto-char start) (not (eobp))) (progn (move-to-left-margin) (not (looking-at parsep))) (or (not (looking-at parstart)) (and use-hard-newlines (not (get-text-property (1- start) 'hard))))) (forward-char 1)) (if (< (point) (point-max)) (goto-char start)))) @end ignore This part handles three situations: when point is between paragraphs, when there is a fill prefix and when there is no fill prefix. @need 800 The @code{while} loop looks like this: @smallexample @group ;; @r{going forwards and not at the end of the buffer} (while (and (> arg 0) (not (eobp))) ;; @r{between paragraphs} ;; Move forward over separator lines... (while (and (not (eobp)) (progn (move-to-left-margin) (not (eobp))) (looking-at parsep)) (forward-line 1)) ;; @r{This decrements the loop} (unless (eobp) (setq arg (1- arg))) ;; ... and one more line. (forward-line 1) @end group @group (if fill-prefix-regexp ;; There is a fill prefix; it overrides parstart; ;; we go forward line by line (while (and (not (eobp)) (progn (move-to-left-margin) (not (eobp))) (not (looking-at parsep)) (looking-at fill-prefix-regexp)) (forward-line 1)) @end group @group ;; There is no fill prefix; ;; we go forward character by character (while (and (re-search-forward sp-parstart nil 1) (progn (setq start (match-beginning 0)) (goto-char start) (not (eobp))) (progn (move-to-left-margin) (not (looking-at parsep))) (or (not (looking-at parstart)) (and use-hard-newlines (not (get-text-property (1- start) 'hard))))) (forward-char 1)) @end group @group ;; and if there is no fill prefix and if we are not at the end, ;; go to whatever was found in the regular expression search ;; for sp-parstart (if (< (point) (point-max)) (goto-char start)))) @end group @end smallexample @findex eobp We can see that this is a decrementing counter @code{while} loop, using the expression @code{(setq arg (1- arg))} as the decrementer. That expression is not far from the @code{while}, but is hidden in another Lisp macro, an @code{unless} macro. Unless we are at the end of the buffer---that is what the @code{eobp} function determines; it is an abbreviation of @samp{End Of Buffer P}---we decrease the value of @code{arg} by one. (If we are at the end of the buffer, we cannot go forward any more and the next loop of the @code{while} expression will test false since the test is an @code{and} with @code{(not (eobp))}. The @code{not} function means exactly as you expect; it is another name for @code{null}, a function that returns true when its argument is false.) Interestingly, the loop count is not decremented until we leave the space between paragraphs, unless we come to the end of buffer or stop seeing the local value of the paragraph separator. That second @code{while} also has a @code{(move-to-left-margin)} expression. The function is self-explanatory. It is inside a @code{progn} expression and not the last element of its body, so it is only invoked for its side effect, which is to move point to the left margin of the current line. @findex looking-at The @code{looking-at} function is also self-explanatory; it returns true if the text after point matches the regular expression given as its argument. The rest of the body of the loop looks difficult at first, but makes sense as you come to understand it. @need 800 First consider what happens if there is a fill prefix: @smallexample @group (if fill-prefix-regexp ;; There is a fill prefix; it overrides parstart; ;; we go forward line by line (while (and (not (eobp)) (progn (move-to-left-margin) (not (eobp))) (not (looking-at parsep)) (looking-at fill-prefix-regexp)) (forward-line 1)) @end group @end smallexample @noindent This expression moves point forward line by line so long as four conditions are true: @enumerate @item Point is not at the end of the buffer. @item We can move to the left margin of the text and are not at the end of the buffer. @item The text following point does not separate paragraphs. @item The pattern following point is the fill prefix regular expression. @end enumerate The last condition may be puzzling, until you remember that point was moved to the beginning of the line early in the @code{forward-paragraph} function. This means that if the text has a fill prefix, the @code{looking-at} function will see it. @need 1250 Consider what happens when there is no fill prefix. @smallexample @group (while (and (re-search-forward sp-parstart nil 1) (progn (setq start (match-beginning 0)) (goto-char start) (not (eobp))) (progn (move-to-left-margin) (not (looking-at parsep))) (or (not (looking-at parstart)) (and use-hard-newlines (not (get-text-property (1- start) 'hard))))) (forward-char 1)) @end group @end smallexample @noindent This @code{while} loop has us searching forward for @code{sp-parstart}, which is the combination of possible whitespace with the local value of the start of a paragraph or of a paragraph separator. (The latter two are within an expression starting @code{\(?:} so that they are not referenced by the @code{match-beginning} function.) @need 800 The two expressions, @smallexample @group (setq start (match-beginning 0)) (goto-char start) @end group @end smallexample @noindent mean go to the start of the text matched by the regular expression search. The @code{(match-beginning 0)} expression is new. It returns a number specifying the location of the start of the text that was matched by the last search. The @code{match-beginning} function is used here because of a characteristic of a forward search: a successful forward search, regardless of whether it is a plain search or a regular expression search, moves point to the end of the text that is found. In this case, a successful search moves point to the end of the pattern for @code{sp-parstart}. However, we want to put point at the end of the current paragraph, not somewhere else. Indeed, since the search possibly includes the paragraph separator, point may end up at the beginning of the next one unless we use an expression that includes @code{match-beginning}. @findex match-beginning When given an argument of 0, @code{match-beginning} returns the position that is the start of the text matched by the most recent search. In this case, the most recent search looks for @code{sp-parstart}. The @code{(match-beginning 0)} expression returns the beginning position of that pattern, rather than the end position of that pattern. (Incidentally, when passed a positive number as an argument, the @code{match-beginning} function returns the location of point at that parenthesized expression in the last search unless that parenthesized expression begins with @code{\(?:}. I don't know why @code{\(?:} appears here since the argument is 0.) @need 1250 The last expression when there is no fill prefix is @smallexample @group (if (< (point) (point-max)) (goto-char start)))) @end group @end smallexample @noindent This says that if there is no fill prefix and if we are not at the end, point should move to the beginning of whatever was found by the regular expression search for @code{sp-parstart}. The full definition for the @code{forward-paragraph} function not only includes code for going forwards, but also code for going backwards. If you are reading this inside of GNU Emacs and you want to see the whole function, you can type @kbd{C-h f} (@code{describe-function}) and the name of the function. This gives you the function documentation and the name of the library containing the function's source. Place point over the name of the library and press the @key{RET} key; you will be taken directly to the source. (Be sure to install your sources! Without them, you are like a person who tries to drive a car with his eyes shut!) @node Regexp Review @section Review Here is a brief summary of some recently introduced functions. @table @code @item while Repeatedly evaluate the body of the expression so long as the first element of the body tests true. Then return @code{nil}. (The expression is evaluated only for its side effects.) @need 1250 For example: @smallexample @group (let ((foo 2)) (while (> foo 0) (insert (format "foo is %d.\n" foo)) (setq foo (1- foo)))) @result{} foo is 2. foo is 1. nil @end group @end smallexample @noindent (The @code{insert} function inserts its arguments at point; the @code{format} function returns a string formatted from its arguments the way @code{message} formats its arguments; @code{\n} produces a new line.) @item re-search-forward Search for a pattern, and if the pattern is found, move point to rest just after it. @noindent Takes four arguments, like @code{search-forward}: @enumerate @item A regular expression that specifies the pattern to search for. (Remember to put quotation marks around this argument!) @item Optionally, the limit of the search. @item Optionally, what to do if the search fails, return @code{nil} or an error message. @item Optionally, how many times to repeat the search; if negative, the search goes backwards. @end enumerate @item let* Bind some variables locally to particular values, and then evaluate the remaining arguments, returning the value of the last one. While binding the local variables, use the local values of variables bound earlier, if any. @need 1250 For example: @smallexample @group (let* ((foo 7) (bar (* 3 foo))) (message "`bar' is %d." bar)) @result{} ‘bar’ is 21. @end group @end smallexample @item match-beginning Return the position of the start of the text found by the last regular expression search. @item looking-at Return @code{t} for true if the text after point matches the argument, which should be a regular expression. @item eobp Return @code{t} for true if point is at the end of the accessible part of a buffer. The end of the accessible part is the end of the buffer if the buffer is not narrowed; it is the end of the narrowed part if the buffer is narrowed. @end table @need 1500 @node re-search Exercises @section Exercises with @code{re-search-forward} @itemize @bullet @item Write a function to search for a regular expression that matches two or more blank lines in sequence. @item Write a function to search for duplicated words, such as ``the the''. @xref{Regexps, , Syntax of Regular Expressions, emacs, The GNU Emacs Manual}, for information on how to write a regexp (a regular expression) to match a string that is composed of two identical halves. You can devise several regexps; some are better than others. The function I use is described in an appendix, along with several regexps. @xref{the-the, , @code{the-the} Duplicated Words Function}. @end itemize @node Counting Words @chapter Counting via Repetition and Regexps @cindex Repetition for word counting @cindex Regular expressions for word counting Repetition and regular expression searches are powerful tools that you often use when you write code in Emacs Lisp. This chapter illustrates the use of regular expression searches through the construction of word count commands using @code{while} loops and recursion. @menu * Why Count Words:: * @value{COUNT-WORDS}:: Use a regexp, but find a problem. * recursive-count-words:: Start with case of no words in region. * Counting Exercise:: @end menu @ifnottex @node Why Count Words @unnumberedsec Counting words @end ifnottex The standard Emacs distribution contains functions for counting the number of lines and words within a region. Certain types of writing ask you to count words. Thus, if you write an essay, you may be limited to 800 words; if you write a novel, you may discipline yourself to write 1000 words a day. It seems odd, but for a long time, Emacs lacked a word count command. Perhaps people used Emacs mostly for code or types of documentation that did not require word counts; or perhaps they restricted themselves to the operating system word count command, @code{wc}. Alternatively, people may have followed the publishers' convention and computed a word count by dividing the number of characters in a document by five. There are many ways to implement a command to count words. Here are some examples, which you may wish to compare with the standard Emacs command, @code{count-words-region}. @node @value{COUNT-WORDS} @section The @code{@value{COUNT-WORDS}} Function @findex @value{COUNT-WORDS} A word count command could count words in a line, paragraph, region, or buffer. What should the command cover? You could design the command to count the number of words in a complete buffer. However, the Emacs tradition encourages flexibility---you may want to count words in just a section, rather than all of a buffer. So it makes more sense to design the command to count the number of words in a region. Once you have a command to count words in a region, you can, if you wish, count words in a whole buffer by marking it with @w{@kbd{C-x h}} (@code{mark-whole-buffer}). Clearly, counting words is a repetitive act: starting from the beginning of the region, you count the first word, then the second word, then the third word, and so on, until you reach the end of the region. This means that word counting is ideally suited to recursion or to a @code{while} loop. @menu * Design @value{COUNT-WORDS}:: The definition using a @code{while} loop. * Whitespace Bug:: The Whitespace Bug in @code{@value{COUNT-WORDS}}. @end menu @ifnottex @node Design @value{COUNT-WORDS} @unnumberedsubsec Designing @code{@value{COUNT-WORDS}} @end ifnottex First, we will implement the word count command with a @code{while} loop, then with recursion. The command will, of course, be interactive. @need 800 The template for an interactive function definition is, as always: @smallexample @group (defun @var{name-of-function} (@var{argument-list}) "@var{documentation}@dots{}" (@var{interactive-expression}@dots{}) @var{body}@dots{}) @end group @end smallexample What we need to do is fill in the slots. The name of the function should be self-explanatory and similar to the existing @code{count-lines-region} name. This makes the name easier to remember. @code{count-words-region} is the obvious choice. Since that name is now used for the standard Emacs command to count words, we will name our implementation @code{@value{COUNT-WORDS}}. The function counts words within a region. This means that the argument list must contain symbols that are bound to the two positions, the beginning and end of the region. These two positions can be called @samp{beginning} and @samp{end} respectively. The first line of the documentation should be a single sentence, since that is all that is printed as documentation by a command such as @code{apropos}. The interactive expression will be of the form @samp{(interactive "r")}, since that will cause Emacs to pass the beginning and end of the region to the function's argument list. All this is routine. The body of the function needs to be written to do three tasks: first, to set up conditions under which the @code{while} loop can count words, second, to run the @code{while} loop, and third, to send a message to the user. When a user calls @code{@value{COUNT-WORDS}}, point may be at the beginning or the end of the region. However, the counting process must start at the beginning of the region. This means we will want to put point there if it is not already there. Executing @code{(goto-char beginning)} ensures this. Of course, we will want to return point to its expected position when the function finishes its work. For this reason, the body must be enclosed in a @code{save-excursion} expression. The central part of the body of the function consists of a @code{while} loop in which one expression jumps point forward word by word, and another expression counts those jumps. The true-or-false-test of the @code{while} loop should test true so long as point should jump forward, and false when point is at the end of the region. We could use @code{(forward-word 1)} as the expression for moving point forward word by word, but it is easier to see what Emacs identifies as a ``word'' if we use a regular expression search. A regular expression search that finds the pattern for which it is searching leaves point after the last character matched. This means that a succession of successful word searches will move point forward word by word. As a practical matter, we want the regular expression search to jump over whitespace and punctuation between words as well as over the words themselves. A regexp that refuses to jump over interword whitespace would never jump more than one word! This means that the regexp should include the whitespace and punctuation that follows a word, if any, as well as the word itself. (A word may end a buffer and not have any following whitespace or punctuation, so that part of the regexp must be optional.) Thus, what we want for the regexp is a pattern defining one or more word constituent characters followed, optionally, by one or more characters that are not word constituents. The regular expression for this is: @smallexample \w+\W* @end smallexample @noindent The buffer's syntax table determines which characters are and are not word constituents. For more information about syntax, @pxref{Syntax Tables, , Syntax Tables, elisp, The GNU Emacs Lisp Reference Manual}. @need 800 The search expression looks like this: @smallexample (re-search-forward "\\w+\\W*") @end smallexample @noindent (Note that paired backslashes precede the @samp{w} and @samp{W}. A single backslash has special meaning to the Emacs Lisp interpreter. It indicates that the following character is interpreted differently than usual. For example, the two characters, @samp{\n}, stand for @samp{newline}, rather than for a backslash followed by @samp{n}. Two backslashes in a row stand for an ordinary, unspecial backslash, so Emacs Lisp interpreter ends of seeing a single backslash followed by a letter. So it discovers the letter is special.) We need a counter to count how many words there are; this variable must first be set to 0 and then incremented each time Emacs goes around the @code{while} loop. The incrementing expression is simply: @smallexample (setq count (1+ count)) @end smallexample Finally, we want to tell the user how many words there are in the region. The @code{message} function is intended for presenting this kind of information to the user. The message has to be phrased so that it reads properly regardless of how many words there are in the region: we don't want to say that ``there are 1 words in the region''. The conflict between singular and plural is ungrammatical. We can solve this problem by using a conditional expression that evaluates different messages depending on the number of words in the region. There are three possibilities: no words in the region, one word in the region, and more than one word. This means that the @code{cond} special form is appropriate. @need 1500 All this leads to the following function definition: @smallexample @group ;;; @r{First version; has bugs!} (defun @value{COUNT-WORDS} (beginning end) "Print number of words in the region. Words are defined as at least one word-constituent character followed by at least one character that is not a word-constituent. The buffer's syntax table determines which characters these are." (interactive "r") (message "Counting words in region ... ") @end group @group ;;; @r{1. Set up appropriate conditions.} (save-excursion (goto-char beginning) (let ((count 0)) @end group @group ;;; @r{2. Run the} while @r{loop.} (while (< (point) end) (re-search-forward "\\w+\\W*") (setq count (1+ count))) @end group @group ;;; @r{3. Send a message to the user.} (cond ((zerop count) (message "The region does NOT have any words.")) ((= 1 count) (message "The region has 1 word.")) (t (message "The region has %d words." count)))))) @end group @end smallexample @noindent As written, the function works, but not in all circumstances. @node Whitespace Bug @subsection The Whitespace Bug in @code{@value{COUNT-WORDS}} The @code{@value{COUNT-WORDS}} command described in the preceding section has two bugs, or rather, one bug with two manifestations. First, if you mark a region containing only whitespace in the middle of some text, the @code{@value{COUNT-WORDS}} command tells you that the region contains one word! Second, if you mark a region containing only whitespace at the end of the buffer or the accessible portion of a narrowed buffer, the command displays an error message that looks like this: @smallexample Search failed: "\\w+\\W*" @end smallexample If you are reading this in Info in GNU Emacs, you can test for these bugs yourself. First, evaluate the function in the usual manner to install it. @ifinfo Here is a copy of the definition. Place your cursor after the closing parenthesis and type @kbd{C-x C-e} to install it. @smallexample @group ;; @r{First version; has bugs!} (defun @value{COUNT-WORDS} (beginning end) "Print number of words in the region. Words are defined as at least one word-constituent character followed by at least one character that is not a word-constituent. The buffer's syntax table determines which characters these are." @end group @group (interactive "r") (message "Counting words in region ... ") @end group @group ;;; @r{1. Set up appropriate conditions.} (save-excursion (goto-char beginning) (let ((count 0)) @end group @group ;;; @r{2. Run the} while @r{loop.} (while (< (point) end) (re-search-forward "\\w+\\W*") (setq count (1+ count))) @end group @group ;;; @r{3. Send a message to the user.} (cond ((zerop count) (message "The region does NOT have any words.")) ((= 1 count) (message "The region has 1 word.")) (t (message "The region has %d words." count)))))) @end group @end smallexample @end ifinfo @need 1000 If you wish, you can also install this keybinding by evaluating it: @smallexample (global-set-key "\C-c=" '@value{COUNT-WORDS}) @end smallexample To conduct the first test, set mark and point to the beginning and end of the following line and then type @kbd{C-c =} (or @kbd{M-x @value{COUNT-WORDS}} if you have not bound @kbd{C-c =}): @smallexample one two three @end smallexample @noindent Emacs will tell you, correctly, that the region has three words. Repeat the test, but place mark at the beginning of the line and place point just @emph{before} the word @samp{one}. Again type the command @kbd{C-c =} (or @kbd{M-x @value{COUNT-WORDS}}). Emacs should tell you that the region has no words, since it is composed only of the whitespace at the beginning of the line. But instead Emacs tells you that the region has one word! For the third test, copy the sample line to the end of the @file{*scratch*} buffer and then type several spaces at the end of the line. Place mark right after the word @samp{three} and point at the end of line. (The end of the line will be the end of the buffer.) Type @kbd{C-c =} (or @kbd{M-x @value{COUNT-WORDS}}) as you did before. Again, Emacs should tell you that the region has no words, since it is composed only of the whitespace at the end of the line. Instead, Emacs displays an error message saying @samp{Search failed}. The two bugs stem from the same problem. Consider the first manifestation of the bug, in which the command tells you that the whitespace at the beginning of the line contains one word. What happens is this: The @code{M-x @value{COUNT-WORDS}} command moves point to the beginning of the region. The @code{while} tests whether the value of point is smaller than the value of @code{end}, which it is. Consequently, the regular expression search looks for and finds the first word. It leaves point after the word. @code{count} is set to one. The @code{while} loop repeats; but this time the value of point is larger than the value of @code{end}, the loop is exited; and the function displays a message saying the number of words in the region is one. In brief, the regular expression search looks for and finds the word even though it is outside the marked region. In the second manifestation of the bug, the region is whitespace at the end of the buffer. Emacs says @samp{Search failed}. What happens is that the true-or-false-test in the @code{while} loop tests true, so the search expression is executed. But since there are no more words in the buffer, the search fails. In both manifestations of the bug, the search extends or attempts to extend outside of the region. The solution is to limit the search to the region---this is a fairly simple action, but as you may have come to expect, it is not quite as simple as you might think. As we have seen, the @code{re-search-forward} function takes a search pattern as its first argument. But in addition to this first, mandatory argument, it accepts three optional arguments. The optional second argument bounds the search. The optional third argument, if @code{t}, causes the function to return @code{nil} rather than signal an error if the search fails. The optional fourth argument is a repeat count. (In Emacs, you can see a function's documentation by typing @kbd{C-h f}, the name of the function, and then @key{RET}.) In the @code{@value{COUNT-WORDS}} definition, the value of the end of the region is held by the variable @code{end} which is passed as an argument to the function. Thus, we can add @code{end} as an argument to the regular expression search expression: @smallexample (re-search-forward "\\w+\\W*" end) @end smallexample However, if you make only this change to the @code{@value{COUNT-WORDS}} definition and then test the new version of the definition on a stretch of whitespace, you will receive an error message saying @samp{Search failed}. What happens is this: the search is limited to the region, and fails as you expect because there are no word-constituent characters in the region. Since it fails, we receive an error message. But we do not want to receive an error message in this case; we want to receive the message ``The region does NOT have any words.'' The solution to this problem is to provide @code{re-search-forward} with a third argument of @code{t}, which causes the function to return @code{nil} rather than signal an error if the search fails. However, if you make this change and try it, you will see the message ``Counting words in region ... '' and @dots{} you will keep on seeing that message @dots{}, until you type @kbd{C-g} (@code{keyboard-quit}). Here is what happens: the search is limited to the region, as before, and it fails because there are no word-constituent characters in the region, as expected. Consequently, the @code{re-search-forward} expression returns @code{nil}. It does nothing else. In particular, it does not move point, which it does as a side effect if it finds the search target. After the @code{re-search-forward} expression returns @code{nil}, the next expression in the @code{while} loop is evaluated. This expression increments the count. Then the loop repeats. The true-or-false-test tests true because the value of point is still less than the value of end, since the @code{re-search-forward} expression did not move point. @dots{} and the cycle repeats @dots{} The @code{@value{COUNT-WORDS}} definition requires yet another modification, to cause the true-or-false-test of the @code{while} loop to test false if the search fails. Put another way, there are two conditions that must be satisfied in the true-or-false-test before the word count variable is incremented: point must still be within the region and the search expression must have found a word to count. Since both the first condition and the second condition must be true together, the two expressions, the region test and the search expression, can be joined with an @code{and} special form and embedded in the @code{while} loop as the true-or-false-test, like this: @smallexample (and (< (point) end) (re-search-forward "\\w+\\W*" end t)) @end smallexample @c colon in printed section title causes problem in Info cross reference @c also trouble with an overfull hbox @iftex @noindent (For information about @code{and}, see @ref{kill-new function, , The @code{kill-new} function}.) @end iftex @ifinfo @noindent (@xref{kill-new function, , The @code{kill-new} function}, for information about @code{and}.) @end ifinfo The @code{re-search-forward} expression returns @code{t} if the search succeeds and as a side effect moves point. Consequently, as words are found, point is moved through the region. When the search expression fails to find another word, or when point reaches the end of the region, the true-or-false-test tests false, the @code{while} loop exits, and the @code{@value{COUNT-WORDS}} function displays one or other of its messages. After incorporating these final changes, the @code{@value{COUNT-WORDS}} works without bugs (or at least, without bugs that I have found!). Here is what it looks like: @smallexample @group ;;; @r{Final version:} @code{while} (defun @value{COUNT-WORDS} (beginning end) "Print number of words in the region." (interactive "r") (message "Counting words in region ... ") @end group @group ;;; @r{1. Set up appropriate conditions.} (save-excursion (let ((count 0)) (goto-char beginning) @end group @group ;;; @r{2. Run the} while @r{loop.} (while (and (< (point) end) (re-search-forward "\\w+\\W*" end t)) (setq count (1+ count))) @end group @group ;;; @r{3. Send a message to the user.} (cond ((zerop count) (message "The region does NOT have any words.")) ((= 1 count) (message "The region has 1 word.")) (t (message "The region has %d words." count)))))) @end group @end smallexample @node recursive-count-words @section Count Words Recursively @cindex Count words recursively @cindex Recursively counting words @cindex Words, counted recursively You can write the function for counting words recursively as well as with a @code{while} loop. Let's see how this is done. First, we need to recognize that the @code{@value{COUNT-WORDS}} function has three jobs: it sets up the appropriate conditions for counting to occur; it counts the words in the region; and it sends a message to the user telling how many words there are. If we write a single recursive function to do everything, we will receive a message for every recursive call. If the region contains 13 words, we will receive thirteen messages, one right after the other. We don't want this! Instead, we must write two functions to do the job, one of which (the recursive function) will be used inside of the other. One function will set up the conditions and display the message; the other will return the word count. Let us start with the function that causes the message to be displayed. We can continue to call this @code{@value{COUNT-WORDS}}. This is the function that the user will call. It will be interactive. Indeed, it will be similar to our previous versions of this function, except that it will call @code{recursive-count-words} to determine how many words are in the region. @need 1250 We can readily construct a template for this function, based on our previous versions: @smallexample @group ;; @r{Recursive version; uses regular expression search} (defun @value{COUNT-WORDS} (beginning end) "@var{documentation}@dots{}" (@var{interactive-expression}@dots{}) @end group @group ;;; @r{1. Set up appropriate conditions.} (@var{explanatory message}) (@var{set-up functions}@dots{} @end group @group ;;; @r{2. Count the words.} @var{recursive call} @end group @group ;;; @r{3. Send a message to the user.} @var{message providing word count})) @end group @end smallexample The definition looks straightforward, except that somehow the count returned by the recursive call must be passed to the message displaying the word count. A little thought suggests that this can be done by making use of a @code{let} expression: we can bind a variable in the varlist of a @code{let} expression to the number of words in the region, as returned by the recursive call; and then the @code{cond} expression, using binding, can display the value to the user. Often, one thinks of the binding within a @code{let} expression as somehow secondary to the primary work of a function. But in this case, what you might consider the primary job of the function, counting words, is done within the @code{let} expression. @need 1250 Using @code{let}, the function definition looks like this: @smallexample @group (defun @value{COUNT-WORDS} (beginning end) "Print number of words in the region." (interactive "r") @end group @group ;;; @r{1. Set up appropriate conditions.} (message "Counting words in region ... ") (save-excursion (goto-char beginning) @end group @group ;;; @r{2. Count the words.} (let ((count (recursive-count-words end))) @end group @group ;;; @r{3. Send a message to the user.} (cond ((zerop count) (message "The region does NOT have any words.")) ((= 1 count) (message "The region has 1 word.")) (t (message "The region has %d words." count)))))) @end group @end smallexample Next, we need to write the recursive counting function. A recursive function has at least three parts: the do-again-test, the next-step-expression, and the recursive call. The do-again-test determines whether the function will or will not be called again. Since we are counting words in a region and can use a function that moves point forward for every word, the do-again-test can check whether point is still within the region. The do-again-test should find the value of point and determine whether point is before, at, or after the value of the end of the region. We can use the @code{point} function to locate point. Clearly, we must pass the value of the end of the region to the recursive counting function as an argument. In addition, the do-again-test should also test whether the search finds a word. If it does not, the function should not call itself again. The next-step-expression changes a value so that when the recursive function is supposed to stop calling itself, it stops. More precisely, the next-step-expression changes a value so that at the right time, the do-again-test stops the recursive function from calling itself again. In this case, the next-step-expression can be the expression that moves point forward, word by word. The third part of a recursive function is the recursive call. Somewhere, we also need a part that does the work of the function, a part that does the counting. A vital part! @need 1250 But already, we have an outline of the recursive counting function: @smallexample @group (defun recursive-count-words (region-end) "@var{documentation}@dots{}" @var{do-again-test} @var{next-step-expression} @var{recursive call}) @end group @end smallexample Now we need to fill in the slots. Let's start with the simplest cases first: if point is at or beyond the end of the region, there cannot be any words in the region, so the function should return zero. Likewise, if the search fails, there are no words to count, so the function should return zero. On the other hand, if point is within the region and the search succeeds, the function should call itself again. @need 800 Thus, the do-again-test should look like this: @smallexample @group (and (< (point) region-end) (re-search-forward "\\w+\\W*" region-end t)) @end group @end smallexample Note that the search expression is part of the do-again-test---the function returns @code{t} if its search succeeds and @code{nil} if it fails. (@xref{Whitespace Bug, , The Whitespace Bug in @code{@value{COUNT-WORDS}}}, for an explanation of how @code{re-search-forward} works.) The do-again-test is the true-or-false test of an @code{if} clause. Clearly, if the do-again-test succeeds, the then-part of the @code{if} clause should call the function again; but if it fails, the else-part should return zero since either point is outside the region or the search failed because there were no words to find. But before considering the recursive call, we need to consider the next-step-expression. What is it? Interestingly, it is the search part of the do-again-test. In addition to returning @code{t} or @code{nil} for the do-again-test, @code{re-search-forward} moves point forward as a side effect of a successful search. This is the action that changes the value of point so that the recursive function stops calling itself when point completes its movement through the region. Consequently, the @code{re-search-forward} expression is the next-step-expression. @need 1200 In outline, then, the body of the @code{recursive-count-words} function looks like this: @smallexample @group (if @var{do-again-test-and-next-step-combined} ;; @r{then} @var{recursive-call-returning-count} ;; @r{else} @var{return-zero}) @end group @end smallexample How to incorporate the mechanism that counts? If you are not used to writing recursive functions, a question like this can be troublesome. But it can and should be approached systematically. We know that the counting mechanism should be associated in some way with the recursive call. Indeed, since the next-step-expression moves point forward by one word, and since a recursive call is made for each word, the counting mechanism must be an expression that adds one to the value returned by a call to @code{recursive-count-words}. @need 800 Consider several cases: @itemize @bullet @item If there are two words in the region, the function should return a value resulting from adding one to the value returned when it counts the first word, plus the number returned when it counts the remaining words in the region, which in this case is one. @item If there is one word in the region, the function should return a value resulting from adding one to the value returned when it counts that word, plus the number returned when it counts the remaining words in the region, which in this case is zero. @item If there are no words in the region, the function should return zero. @end itemize From the sketch we can see that the else-part of the @code{if} returns zero for the case of no words. This means that the then-part of the @code{if} must return a value resulting from adding one to the value returned from a count of the remaining words. @need 1200 The expression will look like this, where @code{1+} is a function that adds one to its argument. @smallexample (1+ (recursive-count-words region-end)) @end smallexample @need 1200 The whole @code{recursive-count-words} function will then look like this: @smallexample @group (defun recursive-count-words (region-end) "@var{documentation}@dots{}" ;;; @r{1. do-again-test} (if (and (< (point) region-end) (re-search-forward "\\w+\\W*" region-end t)) @end group @group ;;; @r{2. then-part: the recursive call} (1+ (recursive-count-words region-end)) ;;; @r{3. else-part} 0)) @end group @end smallexample @need 1250 Let's examine how this works: If there are no words in the region, the else part of the @code{if} expression is evaluated and consequently the function returns zero. If there is one word in the region, the value of point is less than the value of @code{region-end} and the search succeeds. In this case, the true-or-false-test of the @code{if} expression tests true, and the then-part of the @code{if} expression is evaluated. The counting expression is evaluated. This expression returns a value (which will be the value returned by the whole function) that is the sum of one added to the value returned by a recursive call. Meanwhile, the next-step-expression has caused point to jump over the first (and in this case only) word in the region. This means that when @code{(recursive-count-words region-end)} is evaluated a second time, as a result of the recursive call, the value of point will be equal to or greater than the value of region end. So this time, @code{recursive-count-words} will return zero. The zero will be added to one, and the original evaluation of @code{recursive-count-words} will return one plus zero, which is one, which is the correct amount. Clearly, if there are two words in the region, the first call to @code{recursive-count-words} returns one added to the value returned by calling @code{recursive-count-words} on a region containing the remaining word---that is, it adds one to one, producing two, which is the correct amount. Similarly, if there are three words in the region, the first call to @code{recursive-count-words} returns one added to the value returned by calling @code{recursive-count-words} on a region containing the remaining two words---and so on and so on. @need 1250 @noindent With full documentation the two functions look like this: @need 1250 @noindent The recursive function: @findex recursive-count-words @smallexample @group (defun recursive-count-words (region-end) "Number of words between point and REGION-END." @end group @group ;;; @r{1. do-again-test} (if (and (< (point) region-end) (re-search-forward "\\w+\\W*" region-end t)) @end group @group ;;; @r{2. then-part: the recursive call} (1+ (recursive-count-words region-end)) ;;; @r{3. else-part} 0)) @end group @end smallexample @need 800 @noindent The wrapper: @smallexample @group ;;; @r{Recursive version} (defun @value{COUNT-WORDS} (beginning end) "Print number of words in the region. @end group @group Words are defined as at least one word-constituent character followed by at least one character that is not a word-constituent. The buffer's syntax table determines which characters these are." @end group @group (interactive "r") (message "Counting words in region ... ") (save-excursion (goto-char beginning) (let ((count (recursive-count-words end))) @end group @group (cond ((zerop count) (message "The region does NOT have any words.")) @end group @group ((= 1 count) (message "The region has 1 word.")) (t (message "The region has %d words." count)))))) @end group @end smallexample @node Counting Exercise @section Exercise: Counting Punctuation Using a @code{while} loop, write a function to count the number of punctuation marks in a region---period, comma, semicolon, colon, exclamation mark, and question mark. Do the same using recursion. @node Words in a defun @chapter Counting Words in a @code{defun} @cindex Counting words in a @code{defun} @cindex Word counting in a @code{defun} Our next project is to count the number of words in a function definition. Clearly, this can be done using some variant of @code{@value{COUNT-WORDS}}. @xref{Counting Words, , Counting via Repetition and Regexps}. If we are just going to count the words in one definition, it is easy enough to mark the definition with the @kbd{C-M-h} (@code{mark-defun}) command, and then call @code{@value{COUNT-WORDS}}. However, I am more ambitious: I want to count the words and symbols in every definition in the Emacs sources and then print a graph that shows how many functions there are of each length: how many contain 40 to 49 words or symbols, how many contain 50 to 59 words or symbols, and so on. I have often been curious how long a typical function is, and this will tell. @menu * Divide and Conquer:: * Words and Symbols:: What to count? * Syntax:: What constitutes a word or symbol? * count-words-in-defun:: Very like @code{@value{COUNT-WORDS}}. * Several defuns:: Counting several defuns in a file. * Find a File:: Do you want to look at a file? * lengths-list-file:: A list of the lengths of many definitions. * Several files:: Counting in definitions in different files. * Several files recursively:: Recursively counting in different files. * Prepare the data:: Prepare the data for display in a graph. @end menu @ifnottex @node Divide and Conquer @unnumberedsec Divide and Conquer @end ifnottex Described in one phrase, the histogram project is daunting; but divided into numerous small steps, each of which we can take one at a time, the project becomes less fearsome. Let us consider what the steps must be: @itemize @bullet @item First, write a function to count the words in one definition. This includes the problem of handling symbols as well as words. @item Second, write a function to list the number of words in each function in a file. This function can use the @code{count-words-in-defun} function. @item Third, write a function to list the number of words in each function in each of several files. This entails automatically finding the various files, switching to them, and counting the words in the definitions within them. @item Fourth, write a function to convert the list of numbers that we created in step three to a form that will be suitable for printing as a graph. @item Fifth, write a function to print the results as a graph. @end itemize This is quite a project! But if we take each step slowly, it will not be difficult. @node Words and Symbols @section What to Count? @cindex Words and symbols in defun When we first start thinking about how to count the words in a function definition, the first question is (or ought to be) what are we going to count? When we speak of ``words'' with respect to a Lisp function definition, we are actually speaking, in large part, of symbols. For example, the following @code{multiply-by-seven} function contains the five symbols @code{defun}, @code{multiply-by-seven}, @code{number}, @code{*}, and @code{7}. In addition, in the documentation string, it contains the four words @samp{Multiply}, @samp{NUMBER}, @samp{by}, and @samp{seven}. The symbol @samp{number} is repeated, so the definition contains a total of ten words and symbols. @smallexample @group (defun multiply-by-seven (number) "Multiply NUMBER by seven." (* 7 number)) @end group @end smallexample @noindent However, if we mark the @code{multiply-by-seven} definition with @kbd{C-M-h} (@code{mark-defun}), and then call @code{@value{COUNT-WORDS}} on it, we will find that @code{@value{COUNT-WORDS}} claims the definition has eleven words, not ten! Something is wrong! The problem is twofold: @code{@value{COUNT-WORDS}} does not count the @samp{*} as a word, and it counts the single symbol, @code{multiply-by-seven}, as containing three words. The hyphens are treated as if they were interword spaces rather than intraword connectors: @samp{multiply-by-seven} is counted as if it were written @samp{multiply by seven}. The cause of this confusion is the regular expression search within the @code{@value{COUNT-WORDS}} definition that moves point forward word by word. In the canonical version of @code{@value{COUNT-WORDS}}, the regexp is: @smallexample "\\w+\\W*" @end smallexample @noindent This regular expression is a pattern defining one or more word constituent characters possibly followed by one or more characters that are not word constituents. What is meant by ``word constituent characters'' brings us to the issue of syntax, which is worth a section of its own. @node Syntax @section What Constitutes a Word or Symbol? @cindex Syntax categories and tables Emacs treats different characters as belonging to different @dfn{syntax categories}. For example, the regular expression, @samp{\\w+}, is a pattern specifying one or more @emph{word constituent} characters. Word constituent characters are members of one syntax category. Other syntax categories include the class of punctuation characters, such as the period and the comma, and the class of whitespace characters, such as the blank space and the tab character. (For more information, @pxref{Syntax Tables, , Syntax Tables, elisp, The GNU Emacs Lisp Reference Manual}.) Syntax tables specify which characters belong to which categories. Usually, a hyphen is not specified as a word constituent character. Instead, it is specified as being in the class of characters that are part of symbol names but not words. This means that the @code{@value{COUNT-WORDS}} function treats it in the same way it treats an interword white space, which is why @code{@value{COUNT-WORDS}} counts @samp{multiply-by-seven} as three words. There are two ways to cause Emacs to count @samp{multiply-by-seven} as one symbol: modify the syntax table or modify the regular expression. We could redefine a hyphen as a word constituent character by modifying the syntax table that Emacs keeps for each mode. This action would serve our purpose, except that a hyphen is merely the most common character within symbols that is not typically a word constituent character; there are others, too. Alternatively, we can redefine the regexp used in the @code{@value{COUNT-WORDS}} definition so as to include symbols. This procedure has the merit of clarity, but the task is a little tricky. @need 1200 The first part is simple enough: the pattern must match at least one character that is a word or symbol constituent. Thus: @smallexample "\\(\\w\\|\\s_\\)+" @end smallexample @noindent The @samp{\\(} is the first part of the grouping construct that includes the @samp{\\w} and the @samp{\\s_} as alternatives, separated by the @samp{\\|}. The @samp{\\w} matches any word-constituent character and the @samp{\\s_} matches any character that is part of a symbol name but not a word-constituent character. The @samp{+} following the group indicates that the word or symbol constituent characters must be matched at least once. However, the second part of the regexp is more difficult to design. What we want is to follow the first part with optionally one or more characters that are not constituents of a word or symbol. At first, I thought I could define this with the following: @smallexample "\\(\\W\\|\\S_\\)*" @end smallexample @noindent The upper case @samp{W} and @samp{S} match characters that are @emph{not} word or symbol constituents. Unfortunately, this expression matches any character that is either not a word constituent or not a symbol constituent. This matches any character! I then noticed that every word or symbol in my test region was followed by white space (blank space, tab, or newline). So I tried placing a pattern to match one or more blank spaces after the pattern for one or more word or symbol constituents. This failed, too. Words and symbols are often separated by whitespace, but in actual code parentheses may follow symbols and punctuation may follow words. So finally, I designed a pattern in which the word or symbol constituents are followed optionally by characters that are not white space and then followed optionally by white space. @need 800 Here is the full regular expression: @smallexample "\\(\\w\\|\\s_\\)+[^ \t\n]*[ \t\n]*" @end smallexample @node count-words-in-defun @section The @code{count-words-in-defun} Function @cindex Counting words in a @code{defun} We have seen that there are several ways to write a @code{count-words-region} function. To write a @code{count-words-in-defun}, we need merely adapt one of these versions. The version that uses a @code{while} loop is easy to understand, so I am going to adapt that. Because @code{count-words-in-defun} will be part of a more complex program, it need not be interactive and it need not display a message but just return the count. These considerations simplify the definition a little. On the other hand, @code{count-words-in-defun} will be used within a buffer that contains function definitions. Consequently, it is reasonable to ask that the function determine whether it is called when point is within a function definition, and if it is, to return the count for that definition. This adds complexity to the definition, but saves us from needing to pass arguments to the function. @need 1250 These considerations lead us to prepare the following template: @smallexample @group (defun count-words-in-defun () "@var{documentation}@dots{}" (@var{set up}@dots{} (@var{while loop}@dots{}) @var{return count}) @end group @end smallexample @noindent As usual, our job is to fill in the slots. First, the set up. We are presuming that this function will be called within a buffer containing function definitions. Point will either be within a function definition or not. For @code{count-words-in-defun} to work, point must move to the beginning of the definition, a counter must start at zero, and the counting loop must stop when point reaches the end of the definition. The @code{beginning-of-defun} function searches backwards for an opening delimiter such as a @samp{(} at the beginning of a line, and moves point to that position, or else to the limit of the search. In practice, this means that @code{beginning-of-defun} moves point to the beginning of an enclosing or preceding function definition, or else to the beginning of the buffer. We can use @code{beginning-of-defun} to place point where we wish to start. The @code{while} loop requires a counter to keep track of the words or symbols being counted. A @code{let} expression can be used to create a local variable for this purpose, and bind it to an initial value of zero. The @code{end-of-defun} function works like @code{beginning-of-defun} except that it moves point to the end of the definition. @code{end-of-defun} can be used as part of an expression that determines the position of the end of the definition. The set up for @code{count-words-in-defun} takes shape rapidly: first we move point to the beginning of the definition, then we create a local variable to hold the count, and finally, we record the position of the end of the definition so the @code{while} loop will know when to stop looping. @need 1250 The code looks like this: @smallexample @group (beginning-of-defun) (let ((count 0) (end (save-excursion (end-of-defun) (point)))) @end group @end smallexample @noindent The code is simple. The only slight complication is likely to concern @code{end}: it is bound to the position of the end of the definition by a @code{save-excursion} expression that returns the value of point after @code{end-of-defun} temporarily moves it to the end of the definition. The second part of the @code{count-words-in-defun}, after the set up, is the @code{while} loop. The loop must contain an expression that jumps point forward word by word and symbol by symbol, and another expression that counts the jumps. The true-or-false-test for the @code{while} loop should test true so long as point should jump forward, and false when point is at the end of the definition. We have already redefined the regular expression for this, so the loop is straightforward: @smallexample @group (while (and (< (point) end) (re-search-forward "\\(\\w\\|\\s_\\)+[^ \t\n]*[ \t\n]*" end t)) (setq count (1+ count))) @end group @end smallexample The third part of the function definition returns the count of words and symbols. This part is the last expression within the body of the @code{let} expression, and can be, very simply, the local variable @code{count}, which when evaluated returns the count. @need 1250 Put together, the @code{count-words-in-defun} definition looks like this: @findex count-words-in-defun @smallexample @group (defun count-words-in-defun () "Return the number of words and symbols in a defun." (beginning-of-defun) (let ((count 0) (end (save-excursion (end-of-defun) (point)))) @end group @group (while (and (< (point) end) (re-search-forward "\\(\\w\\|\\s_\\)+[^ \t\n]*[ \t\n]*" end t)) (setq count (1+ count))) count)) @end group @end smallexample How to test this? The function is not interactive, but it is easy to put a wrapper around the function to make it interactive; we can use almost the same code as for the recursive version of @code{@value{COUNT-WORDS}}: @smallexample @group ;;; @r{Interactive version.} (defun count-words-defun () "Number of words and symbols in a function definition." (interactive) (message "Counting words and symbols in function definition ... ") @end group @group (let ((count (count-words-in-defun))) (cond ((zerop count) (message "The definition does NOT have any words or symbols.")) @end group @group ((= 1 count) (message "The definition has 1 word or symbol.")) (t (message "The definition has %d words or symbols." count))))) @end group @end smallexample @need 800 @noindent Let's re-use @kbd{C-c =} as a convenient keybinding: @smallexample (global-set-key "\C-c=" 'count-words-defun) @end smallexample Now we can try out @code{count-words-defun}: install both @code{count-words-in-defun} and @code{count-words-defun}, and set the keybinding, and then place the cursor within the following definition: @smallexample @group (defun multiply-by-seven (number) "Multiply NUMBER by seven." (* 7 number)) @result{} 10 @end group @end smallexample @noindent Success! The definition has 10 words and symbols. The next problem is to count the numbers of words and symbols in several definitions within a single file. @node Several defuns @section Count Several @code{defuns} Within a File A file such as @file{simple.el} may have a hundred or more function definitions within it. Our long term goal is to collect statistics on many files, but as a first step, our immediate goal is to collect statistics on one file. The information will be a series of numbers, each number being the length of a function definition. We can store the numbers in a list. We know that we will want to incorporate the information regarding one file with information about many other files; this means that the function for counting definition lengths within one file need only return the list of lengths. It need not and should not display any messages. The word count commands contain one expression to jump point forward word by word and another expression to count the jumps. The function to return the lengths of definitions can be designed to work the same way, with one expression to jump point forward definition by definition and another expression to construct the lengths' list. This statement of the problem makes it elementary to write the function definition. Clearly, we will start the count at the beginning of the file, so the first command will be @code{(goto-char (point-min))}. Next, we start the @code{while} loop; and the true-or-false test of the loop can be a regular expression search for the next function definition---so long as the search succeeds, point is moved forward and then the body of the loop is evaluated. The body needs an expression that constructs the lengths' list. @code{cons}, the list construction command, can be used to create the list. That is almost all there is to it. @need 800 Here is what this fragment of code looks like: @smallexample @group (goto-char (point-min)) (while (re-search-forward "^(defun" nil t) (setq lengths-list (cons (count-words-in-defun) lengths-list))) @end group @end smallexample What we have left out is the mechanism for finding the file that contains the function definitions. In previous examples, we either used this, the Info file, or we switched back and forth to some other buffer, such as the @file{*scratch*} buffer. Finding a file is a new process that we have not yet discussed. @node Find a File @section Find a File @cindex Find a File To find a file in Emacs, you use the @kbd{C-x C-f} (@code{find-file}) command. This command is almost, but not quite right for the lengths problem. @need 1200 Let's look at the source for @code{find-file}: @smallexample @group (defun find-file (filename) "Edit file FILENAME. Switch to a buffer visiting file FILENAME, creating one if none already exists." (interactive "FFind file: ") (switch-to-buffer (find-file-noselect filename))) @end group @end smallexample @noindent (The most recent version of the @code{find-file} function definition permits you to specify optional wildcards to visit multiple files; that makes the definition more complex and we will not discuss it here, since it is not relevant. You can see its source using either @kbd{M-.} (@code{xref-find-definitions}) or @kbd{C-h f} (@code{describe-function}).) @ignore In Emacs 22 (defun find-file (filename &optional wildcards) "Edit file FILENAME. Switch to a buffer visiting file FILENAME, creating one if none already exists. Interactively, the default if you just type @key{RET} is the current directory, but the visited file name is available through the minibuffer history: type M-n to pull it into the minibuffer. Interactively, or if WILDCARDS is non-nil in a call from Lisp, expand wildcards (if any) and visit multiple files. You can suppress wildcard expansion by setting `find-file-wildcards' to nil. To visit a file without any kind of conversion and without automatically choosing a major mode, use \\[find-file-literally]." (interactive (find-file-read-args "Find file: " nil)) (let ((value (find-file-noselect filename nil nil wildcards))) (if (listp value) (mapcar 'switch-to-buffer (nreverse value)) (switch-to-buffer value)))) @end ignore The definition I am showing possesses short but complete documentation and an interactive specification that prompts you for a file name when you use the command interactively. The body of the definition contains two functions, @code{find-file-noselect} and @code{switch-to-buffer}. According to its documentation as shown by @kbd{C-h f} (the @code{describe-function} command), the @code{find-file-noselect} function reads the named file into a buffer and returns the buffer. (Its most recent version includes an optional @var{wildcards} argument, too, as well as another to read a file literally and another to suppress warning messages. These optional arguments are irrelevant.) However, the @code{find-file-noselect} function does not select the buffer in which it puts the file. Emacs does not switch its attention (or yours if you are using @code{find-file-noselect}) to the selected buffer. That is what @code{switch-to-buffer} does: it switches the buffer to which Emacs attention is directed; and it switches the buffer displayed in the window to the new buffer. We have discussed buffer switching elsewhere. (@xref{Switching Buffers}.) In this histogram project, we do not need to display each file on the screen as the program determines the length of each definition within it. Instead of employing @code{switch-to-buffer}, we can work with @code{set-buffer}, which redirects the attention of the computer program to a different buffer but does not redisplay it on the screen. So instead of calling on @code{find-file} to do the job, we must write our own expression. The task is easy: use @code{find-file-noselect} and @code{set-buffer}. @node lengths-list-file @section @code{lengths-list-file} in Detail The core of the @code{lengths-list-file} function is a @code{while} loop containing a function to move point forward defun by defun, and a function to count the number of words and symbols in each defun. This core must be surrounded by functions that do various other tasks, including finding the file, and ensuring that point starts out at the beginning of the file. The function definition looks like this: @findex lengths-list-file @smallexample @group (defun lengths-list-file (filename) "Return list of definitions' lengths within FILE. The returned list is a list of numbers. Each number is the number of words or symbols in one function definition." @end group @group (message "Working on `%s' ... " filename) (save-excursion (let ((buffer (find-file-noselect filename)) (lengths-list)) (set-buffer buffer) (setq buffer-read-only t) (widen) (goto-char (point-min)) (while (re-search-forward "^(defun" nil t) (setq lengths-list (cons (count-words-in-defun) lengths-list))) (kill-buffer buffer) lengths-list))) @end group @end smallexample @noindent The function is passed one argument, the name of the file on which it will work. It has four lines of documentation, but no interactive specification. Since people worry that a computer is broken if they don't see anything going on, the first line of the body is a message. The next line contains a @code{save-excursion} that returns Emacs's attention to the current buffer when the function completes. This is useful in case you embed this function in another function that presumes point is restored to the original buffer. In the varlist of the @code{let} expression, Emacs finds the file and binds the local variable @code{buffer} to the buffer containing the file. At the same time, Emacs creates @code{lengths-list} as a local variable. Next, Emacs switches its attention to the buffer. In the following line, Emacs makes the buffer read-only. Ideally, this line is not necessary. None of the functions for counting words and symbols in a function definition should change the buffer. Besides, the buffer is not going to be saved, even if it were changed. This line is entirely the consequence of great, perhaps excessive, caution. The reason for the caution is that this function and those it calls work on the sources for Emacs and it is inconvenient if they are inadvertently modified. It goes without saying that I did not realize a need for this line until an experiment went awry and started to modify my Emacs source files @dots{} Next comes a call to widen the buffer if it is narrowed. This function is usually not needed---Emacs creates a fresh buffer if none already exists; but if a buffer visiting the file already exists Emacs returns that one. In this case, the buffer may be narrowed and must be widened. If we wanted to be fully user-friendly, we would arrange to save the restriction and the location of point, but we won't. The @code{(goto-char (point-min))} expression moves point to the beginning of the buffer. Then comes a @code{while} loop in which the work of the function is carried out. In the loop, Emacs determines the length of each definition and constructs a lengths' list containing the information. Emacs kills the buffer after working through it. This is to save space inside of Emacs. My version of GNU Emacs 19 contained over 300 source files of interest; GNU Emacs 22 contains over a thousand source files. Another function will apply @code{lengths-list-file} to each of the files. Finally, the last expression within the @code{let} expression is the @code{lengths-list} variable; its value is returned as the value of the whole function. You can try this function by installing it in the usual fashion. Then place your cursor after the following expression and type @kbd{C-x C-e} (@code{eval-last-sexp}). @c !!! 22.1.1 lisp sources location here @smallexample (lengths-list-file "/usr/local/share/emacs/22.1/lisp/emacs-lisp/debug.el") @end smallexample @noindent You may need to change the pathname of the file; the one here is for GNU Emacs version 22.1. To change the expression, copy it to the @file{*scratch*} buffer and edit it. @need 1200 @noindent Also, to see the full length of the list, rather than a truncated version, you may have to evaluate the following: @c We do not want to insert, so do not mention the zero prefix argument. @smallexample (custom-set-variables '(eval-expression-print-length nil)) @end smallexample @noindent (@xref{defcustom, , Specifying Variables using @code{defcustom}}. Then evaluate the @code{lengths-list-file} expression.) @need 1200 The lengths' list for @file{debug.el} takes less than a second to produce and looks like this in GNU Emacs 22: @smallexample (83 113 105 144 289 22 30 97 48 89 25 52 52 88 28 29 77 49 43 290 232 587) @end smallexample @need 1500 (Using my old machine, the version 19 lengths' list for @file{debug.el} took seven seconds to produce and looked like this: @smallexample (75 41 80 62 20 45 44 68 45 12 34 235) @end smallexample @noindent The newer version of @file{debug.el} contains more defuns than the earlier one; and my new machine is much faster than the old one.) Note that the length of the last definition in the file is first in the list. @node Several files @section Count Words in @code{defuns} in Different Files In the previous section, we created a function that returns a list of the lengths of each definition in a file. Now, we want to define a function to return a master list of the lengths of the definitions in a list of files. Working on each of a list of files is a repetitious act, so we can use either a @code{while} loop or recursion. @menu * lengths-list-many-files:: Return a list of the lengths of defuns. * append:: Attach one list to another. @end menu @ifnottex @node lengths-list-many-files @unnumberedsubsec Determine the lengths of @code{defuns} @end ifnottex The design using a @code{while} loop is routine. The argument passed to the function is a list of files. As we saw earlier (@pxref{Loop Example}), you can write a @code{while} loop so that the body of the loop is evaluated if such a list contains elements, but to exit the loop if the list is empty. For this design to work, the body of the loop must contain an expression that shortens the list each time the body is evaluated, so that eventually the list is empty. The usual technique is to set the value of the list to the value of the @sc{cdr} of the list each time the body is evaluated. @need 800 The template looks like this: @smallexample @group (while @var{test-whether-list-is-empty} @var{body}@dots{} @var{set-list-to-cdr-of-list}) @end group @end smallexample Also, we remember that a @code{while} loop returns @code{nil} (the result of evaluating the true-or-false-test), not the result of any evaluation within its body. (The evaluations within the body of the loop are done for their side effects.) However, the expression that sets the lengths' list is part of the body---and that is the value that we want returned by the function as a whole. To do this, we enclose the @code{while} loop within a @code{let} expression, and arrange that the last element of the @code{let} expression contains the value of the lengths' list. (@xref{Incrementing Example, , Loop Example with an Incrementing Counter}.) @findex lengths-list-many-files @need 1250 These considerations lead us directly to the function itself: @smallexample @group ;;; @r{Use @code{while} loop.} (defun lengths-list-many-files (list-of-files) "Return list of lengths of defuns in LIST-OF-FILES." @end group @group (let (lengths-list) ;;; @r{true-or-false-test} (while list-of-files (setq lengths-list (append lengths-list ;;; @r{Generate a lengths' list.} (lengths-list-file (expand-file-name (car list-of-files))))) @end group @group ;;; @r{Make files' list shorter.} (setq list-of-files (cdr list-of-files))) ;;; @r{Return final value of lengths' list.} lengths-list)) @end group @end smallexample @code{expand-file-name} is a built-in function that converts a file name to the absolute, long, path name form. The function employs the name of the directory in which the function is called. @c !!! 22.1.1 lisp sources location here @need 1500 Thus, if @code{expand-file-name} is called on @code{debug.el} when Emacs is visiting the @file{/usr/local/share/emacs/22.1.1/lisp/emacs-lisp/} directory, @smallexample debug.el @end smallexample @need 800 @noindent becomes @c !!! 22.1.1 lisp sources location here @smallexample /usr/local/share/emacs/22.1.1/lisp/emacs-lisp/debug.el @end smallexample The only other new element of this function definition is the as yet unstudied function @code{append}, which merits a short section for itself. @node append @subsection The @code{append} Function @need 800 The @code{append} function attaches one list to another. Thus, @smallexample (append '(1 2 3 4) '(5 6 7 8)) @end smallexample @need 800 @noindent produces the list @smallexample (1 2 3 4 5 6 7 8) @end smallexample This is exactly how we want to attach two lengths' lists produced by @code{lengths-list-file} to each other. The results contrast with @code{cons}, @smallexample (cons '(1 2 3 4) '(5 6 7 8)) @end smallexample @need 1250 @noindent which constructs a new list in which the first argument to @code{cons} becomes the first element of the new list: @smallexample ((1 2 3 4) 5 6 7 8) @end smallexample @node Several files recursively @section Recursively Count Words in Different Files Besides a @code{while} loop, you can work on each of a list of files with recursion. A recursive version of @code{lengths-list-many-files} is short and simple. The recursive function has the usual parts: the do-again-test, the next-step-expression, and the recursive call. The do-again-test determines whether the function should call itself again, which it will do if the @code{list-of-files} contains any remaining elements; the next-step-expression resets the @code{list-of-files} to the @sc{cdr} of itself, so eventually the list will be empty; and the recursive call calls itself on the shorter list. The complete function is shorter than this description! @findex recursive-lengths-list-many-files @smallexample @group (defun recursive-lengths-list-many-files (list-of-files) "Return list of lengths of each defun in LIST-OF-FILES." (if list-of-files ; @r{do-again-test} (append (lengths-list-file (expand-file-name (car list-of-files))) (recursive-lengths-list-many-files (cdr list-of-files))))) @end group @end smallexample @noindent In a sentence, the function returns the lengths' list for the first of the @code{list-of-files} appended to the result of calling itself on the rest of the @code{list-of-files}. Here is a test of @code{recursive-lengths-list-many-files}, along with the results of running @code{lengths-list-file} on each of the files individually. Install @code{recursive-lengths-list-many-files} and @code{lengths-list-file}, if necessary, and then evaluate the following expressions. You may need to change the files' pathnames; those here work when this Info file and the Emacs sources are located in their customary places. To change the expressions, copy them to the @file{*scratch*} buffer, edit them, and then evaluate them. The results are shown after the @samp{@result{}}. (These results are for files from Emacs version 22.1.1; files from other versions of Emacs may produce different results.) @c !!! 22.1.1 lisp sources location here @smallexample @group (cd "/usr/local/share/emacs/22.1.1/") (lengths-list-file "./lisp/macros.el") @result{} (283 263 480 90) @end group @group (lengths-list-file "./lisp/mail/mailalias.el") @result{} (38 32 29 95 178 180 321 218 324) @end group @group (lengths-list-file "./lisp/makesum.el") @result{} (85 181) @end group @group (recursive-lengths-list-many-files '("./lisp/macros.el" "./lisp/mail/mailalias.el" "./lisp/makesum.el")) @result{} (283 263 480 90 38 32 29 95 178 180 321 218 324 85 181) @end group @end smallexample The @code{recursive-lengths-list-many-files} function produces the output we want. The next step is to prepare the data in the list for display in a graph. @node Prepare the data @section Prepare the Data for Display in a Graph The @code{recursive-lengths-list-many-files} function returns a list of numbers. Each number records the length of a function definition. What we need to do now is transform this data into a list of numbers suitable for generating a graph. The new list will tell how many functions definitions contain less than 10 words and symbols, how many contain between 10 and 19 words and symbols, how many contain between 20 and 29 words and symbols, and so on. In brief, we need to go through the lengths' list produced by the @code{recursive-lengths-list-many-files} function and count the number of defuns within each range of lengths, and produce a list of those numbers. @menu * Data for Display in Detail:: * Sorting:: Sorting lists. * Files List:: Making a list of files. * Counting function definitions:: @end menu @ifnottex @node Data for Display in Detail @unnumberedsubsec The Data for Display in Detail @end ifnottex Based on what we have done before, we can readily foresee that it should not be too hard to write a function that @sc{cdr}s down the lengths' list, looks at each element, determines which length range it is in, and increments a counter for that range. However, before beginning to write such a function, we should consider the advantages of sorting the lengths' list first, so the numbers are ordered from smallest to largest. First, sorting will make it easier to count the numbers in each range, since two adjacent numbers will either be in the same length range or in adjacent ranges. Second, by inspecting a sorted list, we can discover the highest and lowest number, and thereby determine the largest and smallest length range that we will need. @node Sorting @subsection Sorting Lists @findex sort Emacs contains a function to sort lists, called (as you might guess) @code{sort}. The @code{sort} function takes two arguments, the list to be sorted, and a predicate that determines whether the first of two list elements is less than the second. As we saw earlier (@pxref{Wrong Type of Argument, , Using the Wrong Type Object as an Argument}), a predicate is a function that determines whether some property is true or false. The @code{sort} function will reorder a list according to whatever property the predicate uses; this means that @code{sort} can be used to sort non-numeric lists by non-numeric criteria---it can, for example, alphabetize a list. @need 1250 The @code{<} function is used when sorting a numeric list. For example, @smallexample (sort '(4 8 21 17 33 7 21 7) '<) @end smallexample @need 800 @noindent produces this: @smallexample (4 7 7 8 17 21 21 33) @end smallexample @noindent (Note that in this example, both the arguments are quoted so that the symbols are not evaluated before being passed to @code{sort} as arguments.) Sorting the list returned by the @code{recursive-lengths-list-many-files} function is straightforward; it uses the @code{<} function: @ignore 2006 Oct 29 In GNU Emacs 22, eval (progn (cd "/usr/local/share/emacs/22.0.50/") (sort (recursive-lengths-list-many-files '("./lisp/macros.el" "./lisp/mail/mailalias.el" "./lisp/makesum.el")) '<)) @end ignore @smallexample @group (sort (recursive-lengths-list-many-files '("./lisp/macros.el" "./lisp/mailalias.el" "./lisp/makesum.el")) '<) @end group @end smallexample @need 800 @noindent which produces: @smallexample (29 32 38 85 90 95 178 180 181 218 263 283 321 324 480) @end smallexample @noindent (Note that in this example, the first argument to @code{sort} is not quoted, since the expression must be evaluated so as to produce the list that is passed to @code{sort}.) @node Files List @subsection Making a List of Files The @code{recursive-lengths-list-many-files} function requires a list of files as its argument. For our test examples, we constructed such a list by hand; but the Emacs Lisp source directory is too large for us to do for that. Instead, we will write a function to do the job for us. In this function, we will use both a @code{while} loop and a recursive call. @findex directory-files We did not have to write a function like this for older versions of GNU Emacs, since they placed all the @samp{.el} files in one directory. Instead, we were able to use the @code{directory-files} function, which lists the names of files that match a specified pattern within a single directory. However, recent versions of Emacs place Emacs Lisp files in sub-directories of the top level @file{lisp} directory. This re-arrangement eases navigation. For example, all the mail related files are in a @file{lisp} sub-directory called @file{mail}. But at the same time, this arrangement forces us to create a file listing function that descends into the sub-directories. @findex files-in-below-directory We can create this function, called @code{files-in-below-directory}, using familiar functions such as @code{car}, @code{nthcdr}, and @code{substring} in conjunction with an existing function called @code{directory-files-and-attributes}. This latter function not only lists all the filenames in a directory, including the names of sub-directories, but also their attributes. To restate our goal: to create a function that will enable us to feed filenames to @code{recursive-lengths-list-many-files} as a list that looks like this (but with more elements): @smallexample @group ("./lisp/macros.el" "./lisp/mail/rmail.el" "./lisp/makesum.el") @end group @end smallexample The @code{directory-files-and-attributes} function returns a list of lists. Each of the lists within the main list consists of 13 elements. The first element is a string that contains the name of the file---which, in GNU/Linux, may be a @dfn{directory file}, that is to say, a file with the special attributes of a directory. The second element of the list is @code{t} for a directory, a string for symbolic link (the string is the name linked to), or @code{nil}. For example, the first @samp{.el} file in the @file{lisp/} directory is @file{abbrev.el}. Its name is @file{/usr/local/share/emacs/22.1.1/lisp/abbrev.el} and it is not a directory or a symbolic link. @need 1000 This is how @code{directory-files-and-attributes} lists that file and its attributes: @smallexample @group ("abbrev.el" nil 1 1000 100 @end group @group (20615 27034 579989 697000) (17905 55681 0 0) (20615 26327 734791 805000) 13188 "-rw-r--r--" @end group @group t 2971624 773) @end group @end smallexample @need 1200 On the other hand, @file{mail/} is a directory within the @file{lisp/} directory. The beginning of its listing looks like this: @smallexample @group ("mail" t @dots{} ) @end group @end smallexample (To learn about the different attributes, look at the documentation of @code{file-attributes}. Bear in mind that the @code{file-attributes} function does not list the filename, so its first element is @code{directory-files-and-attributes}'s second element.) We will want our new function, @code{files-in-below-directory}, to list the @samp{.el} files in the directory it is told to check, and in any directories below that directory. This gives us a hint on how to construct @code{files-in-below-directory}: within a directory, the function should add @samp{.el} filenames to a list; and if, within a directory, the function comes upon a sub-directory, it should go into that sub-directory and repeat its actions. However, we should note that every directory contains a name that refers to itself, called @file{.} (``dot''), and a name that refers to its parent directory, called @file{..} (``dot dot''). (In @file{/}, the root directory, @file{..} refers to itself, since @file{/} has no parent.) Clearly, we do not want our @code{files-in-below-directory} function to enter those directories, since they always lead us, directly or indirectly, to the current directory. Consequently, our @code{files-in-below-directory} function must do several tasks: @itemize @bullet @item Check to see whether it is looking at a filename that ends in @samp{.el}; and if so, add its name to a list. @item Check to see whether it is looking at a filename that is the name of a directory; and if so, @itemize @minus @item Check to see whether it is looking at @file{.} or @file{..}; and if so skip it. @item Or else, go into that directory and repeat the process. @end itemize @end itemize Let's write a function definition to do these tasks. We will use a @code{while} loop to move from one filename to another within a directory, checking what needs to be done; and we will use a recursive call to repeat the actions on each sub-directory. The recursive pattern is Accumulate (@pxref{Accumulate}), using @code{append} as the combiner. @ignore (directory-files "/usr/local/src/emacs/lisp/" t "\\.el$") (shell-command "find /usr/local/src/emacs/lisp/ -name '*.el'") (directory-files "/usr/local/share/emacs/22.1.1/lisp/" t "\\.el$") (shell-command "find /usr/local/share/emacs/22.1.1/lisp/ -name '*.el'") @end ignore @c /usr/local/share/emacs/22.1.1/lisp/ @need 800 Here is the function: @smallexample @group (defun files-in-below-directory (directory) "List the .el files in DIRECTORY and in its sub-directories." ;; Although the function will be used non-interactively, ;; it will be easier to test if we make it interactive. ;; The directory will have a name such as ;; "/usr/local/share/emacs/22.1.1/lisp/" (interactive "DDirectory name: ") @end group @group (let (el-files-list (current-directory-list (directory-files-and-attributes directory t))) ;; while we are in the current directory (while current-directory-list @end group @group (cond ;; check to see whether filename ends in '.el' ;; and if so, add its name to a list. ((equal ".el" (substring (car (car current-directory-list)) -3)) (setq el-files-list (cons (car (car current-directory-list)) el-files-list))) @end group @group ;; check whether filename is that of a directory ((eq t (car (cdr (car current-directory-list)))) ;; decide whether to skip or recurse (if (equal "." (substring (car (car current-directory-list)) -1)) ;; then do nothing since filename is that of ;; current directory or parent, "." or ".." () @end group @group ;; else descend into the directory and repeat the process (setq el-files-list (append (files-in-below-directory (car (car current-directory-list))) el-files-list))))) ;; move to the next filename in the list; this also ;; shortens the list so the while loop eventually comes to an end (setq current-directory-list (cdr current-directory-list))) ;; return the filenames el-files-list)) @end group @end smallexample @c (files-in-below-directory "/usr/local/src/emacs/lisp/") @c (files-in-below-directory "/usr/local/share/emacs/22.1.1/lisp/") The @code{files-in-below-directory} @code{directory-files} function takes one argument, the name of a directory. @need 1250 Thus, on my system, @c (length (files-in-below-directory "/usr/local/src/emacs/lisp/")) @c !!! 22.1.1 lisp sources location here @smallexample @group (length (files-in-below-directory "/usr/local/share/emacs/22.1.1/lisp/")) @end group @end smallexample @noindent tells me that in and below my Lisp sources directory are 1031 @samp{.el} files. @code{files-in-below-directory} returns a list in reverse alphabetical order. An expression to sort the list in alphabetical order looks like this: @smallexample @group (sort (files-in-below-directory "/usr/local/share/emacs/22.1.1/lisp/") 'string-lessp) @end group @end smallexample @ignore (defun test () "Test how long it takes to find lengths of all sorted elisp defuns." (insert "\n" (current-time-string) "\n") (sit-for 0) (sort (recursive-lengths-list-many-files (files-in-below-directory "/usr/local/src/emacs/lisp/")) '<) (insert (current-time-string))) @end ignore @node Counting function definitions @subsection Counting function definitions Our immediate goal is to generate a list that tells us how many function definitions contain fewer than 10 words and symbols, how many contain between 10 and 19 words and symbols, how many contain between 20 and 29 words and symbols, and so on. With a sorted list of numbers, this is easy: count how many elements of the list are smaller than 10, then, after moving past the numbers just counted, count how many are smaller than 20, then, after moving past the numbers just counted, count how many are smaller than 30, and so on. Each of the numbers, 10, 20, 30, 40, and the like, is one larger than the top of that range. We can call the list of such numbers the @code{top-of-ranges} list. @need 1200 If we wished, we could generate this list automatically, but it is simpler to write a list manually. Here it is: @vindex top-of-ranges @smallexample @group (defvar top-of-ranges '(10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250 260 270 280 290 300) "List specifying ranges for `defuns-per-range'.") @end group @end smallexample To change the ranges, we edit this list. Next, we need to write the function that creates the list of the number of definitions within each range. Clearly, this function must take the @code{sorted-lengths} and the @code{top-of-ranges} lists as arguments. The @code{defuns-per-range} function must do two things again and again: it must count the number of definitions within a range specified by the current top-of-range value; and it must shift to the next higher value in the @code{top-of-ranges} list after counting the number of definitions in the current range. Since each of these actions is repetitive, we can use @code{while} loops for the job. One loop counts the number of definitions in the range defined by the current top-of-range value, and the other loop selects each of the top-of-range values in turn. Several entries of the @code{sorted-lengths} list are counted for each range; this means that the loop for the @code{sorted-lengths} list will be inside the loop for the @code{top-of-ranges} list, like a small gear inside a big gear. The inner loop counts the number of definitions within the range. It is a simple counting loop of the type we have seen before. (@xref{Incrementing Loop, , A loop with an incrementing counter}.) The true-or-false test of the loop tests whether the value from the @code{sorted-lengths} list is smaller than the current value of the top of the range. If it is, the function increments the counter and tests the next value from the @code{sorted-lengths} list. @need 1250 The inner loop looks like this: @smallexample @group (while @var{length-element-smaller-than-top-of-range} (setq number-within-range (1+ number-within-range)) (setq sorted-lengths (cdr sorted-lengths))) @end group @end smallexample The outer loop must start with the lowest value of the @code{top-of-ranges} list, and then be set to each of the succeeding higher values in turn. This can be done with a loop like this: @smallexample @group (while top-of-ranges @var{body-of-loop}@dots{} (setq top-of-ranges (cdr top-of-ranges))) @end group @end smallexample @need 1200 Put together, the two loops look like this: @smallexample @group (while top-of-ranges ;; @r{Count the number of elements within the current range.} (while @var{length-element-smaller-than-top-of-range} (setq number-within-range (1+ number-within-range)) (setq sorted-lengths (cdr sorted-lengths))) ;; @r{Move to next range.} (setq top-of-ranges (cdr top-of-ranges))) @end group @end smallexample In addition, in each circuit of the outer loop, Emacs should record the number of definitions within that range (the value of @code{number-within-range}) in a list. We can use @code{cons} for this purpose. (@xref{cons, , @code{cons}}.) The @code{cons} function works fine, except that the list it constructs will contain the number of definitions for the highest range at its beginning and the number of definitions for the lowest range at its end. This is because @code{cons} attaches new elements of the list to the beginning of the list, and since the two loops are working their way through the lengths' list from the lower end first, the @code{defuns-per-range-list} will end up largest number first. But we will want to print our graph with smallest values first and the larger later. The solution is to reverse the order of the @code{defuns-per-range-list}. We can do this using the @code{nreverse} function, which reverses the order of a list. @findex nreverse @need 800 For example, @smallexample (nreverse '(1 2 3 4)) @end smallexample @need 800 @noindent produces: @smallexample (4 3 2 1) @end smallexample Note that the @code{nreverse} function is destructive---that is, it changes the list to which it is applied; this contrasts with the @code{car} and @code{cdr} functions, which are non-destructive. In this case, we do not want the original @code{defuns-per-range-list}, so it does not matter that it is destroyed. (The @code{reverse} function provides a reversed copy of a list, leaving the original list as is.) @findex reverse @need 1250 Put all together, the @code{defuns-per-range} looks like this: @smallexample @group (defun defuns-per-range (sorted-lengths top-of-ranges) "SORTED-LENGTHS defuns in each TOP-OF-RANGES range." (let ((top-of-range (car top-of-ranges)) (number-within-range 0) defuns-per-range-list) @end group @group ;; @r{Outer loop.} (while top-of-ranges @end group @group ;; @r{Inner loop.} (while (and ;; @r{Need number for numeric test.} (car sorted-lengths) (< (car sorted-lengths) top-of-range)) @end group @group ;; @r{Count number of definitions within current range.} (setq number-within-range (1+ number-within-range)) (setq sorted-lengths (cdr sorted-lengths))) ;; @r{Exit inner loop but remain within outer loop.} @end group @group (setq defuns-per-range-list (cons number-within-range defuns-per-range-list)) (setq number-within-range 0) ; @r{Reset count to zero.} @end group @group ;; @r{Move to next range.} (setq top-of-ranges (cdr top-of-ranges)) ;; @r{Specify next top of range value.} (setq top-of-range (car top-of-ranges))) @end group @group ;; @r{Exit outer loop and count the number of defuns larger than} ;; @r{ the largest top-of-range value.} (setq defuns-per-range-list (cons (length sorted-lengths) defuns-per-range-list)) @end group @group ;; @r{Return a list of the number of definitions within each range,} ;; @r{ smallest to largest.} (nreverse defuns-per-range-list))) @end group @end smallexample @need 1200 @noindent The function is straightforward except for one subtle feature. The true-or-false test of the inner loop looks like this: @smallexample @group (and (car sorted-lengths) (< (car sorted-lengths) top-of-range)) @end group @end smallexample @need 800 @noindent instead of like this: @smallexample (< (car sorted-lengths) top-of-range) @end smallexample The purpose of the test is to determine whether the first item in the @code{sorted-lengths} list is less than the value of the top of the range. The simple version of the test works fine unless the @code{sorted-lengths} list has a @code{nil} value. In that case, the @code{(car sorted-lengths)} expression function returns @code{nil}. The @code{<} function cannot compare a number to @code{nil}, which is an empty list, so Emacs signals an error and stops the function from attempting to continue to execute. The @code{sorted-lengths} list always becomes @code{nil} when the counter reaches the end of the list. This means that any attempt to use the @code{defuns-per-range} function with the simple version of the test will fail. We solve the problem by using the @code{(car sorted-lengths)} expression in conjunction with the @code{and} expression. The @code{(car sorted-lengths)} expression returns a non-@code{nil} value so long as the list has at least one number within it, but returns @code{nil} if the list is empty. The @code{and} expression first evaluates the @code{(car sorted-lengths)} expression, and if it is @code{nil}, returns false @emph{without} evaluating the @code{<} expression. But if the @code{(car sorted-lengths)} expression returns a non-@code{nil} value, the @code{and} expression evaluates the @code{<} expression, and returns that value as the value of the @code{and} expression. @c colon in printed section title causes problem in Info cross reference This way, we avoid an error. @iftex @noindent (For information about @code{and}, see @ref{kill-new function, , The @code{kill-new} function}.) @end iftex @ifinfo @noindent (@xref{kill-new function, , The @code{kill-new} function}, for information about @code{and}.) @end ifinfo Here is a short test of the @code{defuns-per-range} function. First, evaluate the expression that binds (a shortened) @code{top-of-ranges} list to the list of values, then evaluate the expression for binding the @code{sorted-lengths} list, and then evaluate the @code{defuns-per-range} function. @smallexample @group ;; @r{(Shorter list than we will use later.)} (setq top-of-ranges '(110 120 130 140 150 160 170 180 190 200)) (setq sorted-lengths '(85 86 110 116 122 129 154 176 179 200 265 300 300)) (defuns-per-range sorted-lengths top-of-ranges) @end group @end smallexample @need 800 @noindent The list returned looks like this: @smallexample (2 2 2 0 0 1 0 2 0 0 4) @end smallexample @noindent Indeed, there are two elements of the @code{sorted-lengths} list smaller than 110, two elements between 110 and 119, two elements between 120 and 129, and so on. There are four elements with a value of 200 or larger. @c The next step is to turn this numbers' list into a graph. @node Readying a Graph @chapter Readying a Graph @cindex Readying a graph @cindex Graph prototype @cindex Prototype graph @cindex Body of graph Our goal is to construct a graph showing the numbers of function definitions of various lengths in the Emacs lisp sources. As a practical matter, if you were creating a graph, you would probably use a program such as @code{gnuplot} to do the job. (@code{gnuplot} is nicely integrated into GNU Emacs.) In this case, however, we create one from scratch, and in the process we will re-acquaint ourselves with some of what we learned before and learn more. In this chapter, we will first write a simple graph printing function. This first definition will be a @dfn{prototype}, a rapidly written function that enables us to reconnoiter this unknown graph-making territory. We will discover dragons, or find that they are myth. After scouting the terrain, we will feel more confident and enhance the function to label the axes automatically. @menu * Columns of a graph:: * graph-body-print:: How to print the body of a graph. * recursive-graph-body-print:: * Printed Axes:: * Line Graph Exercise:: @end menu @ifnottex @node Columns of a graph @unnumberedsec Printing the Columns of a Graph @end ifnottex Since Emacs is designed to be flexible and work with all kinds of terminals, including character-only terminals, the graph will need to be made from one of the typewriter symbols. An asterisk will do; as we enhance the graph-printing function, we can make the choice of symbol a user option. We can call this function @code{graph-body-print}; it will take a @code{numbers-list} as its only argument. At this stage, we will not label the graph, but only print its body. The @code{graph-body-print} function inserts a vertical column of asterisks for each element in the @code{numbers-list}. The height of each line is determined by the value of that element of the @code{numbers-list}. Inserting columns is a repetitive act; that means that this function can be written either with a @code{while} loop or recursively. Our first challenge is to discover how to print a column of asterisks. Usually, in Emacs, we print characters onto a screen horizontally, line by line, by typing. We have two routes we can follow: write our own column-insertion function or discover whether one exists in Emacs. To see whether there is one in Emacs, we can use the @kbd{M-x apropos} command. This command is like the @kbd{C-h a} (@code{command-apropos}) command, except that the latter finds only those functions that are commands. The @kbd{M-x apropos} command lists all symbols that match a regular expression, including functions that are not interactive. @findex apropos What we want to look for is some command that prints or inserts columns. Very likely, the name of the function will contain either the word ``print'' or the word ``insert'' or the word ``column''. Therefore, we can simply type @kbd{M-x apropos @key{RET} print\|insert\|column @key{RET}} and look at the result. On my system, this command once took quite some time, and then produced a list of 79 functions and variables. Now it does not take much time at all and produces a list of 211 functions and variables. Scanning down the list, the only function that looks as if it might do the job is @code{insert-rectangle}. @need 1200 Indeed, this is the function we want; its documentation says: @smallexample @group insert-rectangle: Insert text of RECTANGLE with upper left corner at point. RECTANGLE's first line is inserted at point, its second line is inserted at a point vertically under point, etc. RECTANGLE should be a list of strings. After this command, the mark is at the upper left corner and point is at the lower right corner. @end group @end smallexample We can run a quick test, to make sure it does what we expect of it. Here is the result of placing the cursor after the @code{insert-rectangle} expression and typing @kbd{C-u C-x C-e} (@code{eval-last-sexp}). The function inserts the strings @samp{"first"}, @samp{"second"}, and @samp{"third"} at and below point. Also the function returns @code{nil}. @smallexample @group (insert-rectangle '("first" "second" "third"))first second thirdnil @end group @end smallexample @noindent Of course, we won't be inserting the text of the @code{insert-rectangle} expression itself into the buffer in which we are making the graph, but will call the function from our program. We shall, however, have to make sure that point is in the buffer at the place where the @code{insert-rectangle} function will insert its column of strings. If you are reading this in Info, you can see how this works by switching to another buffer, such as the @file{*scratch*} buffer, placing point somewhere in the buffer, typing @kbd{M-:}, typing the @code{insert-rectangle} expression into the minibuffer at the prompt, and then typing @key{RET}. This causes Emacs to evaluate the expression in the minibuffer, but to use as the value of point the position of point in the @file{*scratch*} buffer. (@kbd{M-:} is the keybinding for @code{eval-expression}. Also, @code{nil} does not appear in the @file{*scratch*} buffer since the expression is evaluated in the minibuffer.) We find when we do this that point ends up at the end of the last inserted line---that is to say, this function moves point as a side-effect. If we were to repeat the command, with point at this position, the next insertion would be below and to the right of the previous insertion. We don't want this! If we are going to make a bar graph, the columns need to be beside each other. So we discover that each cycle of the column-inserting @code{while} loop must reposition point to the place we want it, and that place will be at the top, not the bottom, of the column. Moreover, we remember that when we print a graph, we do not expect all the columns to be the same height. This means that the top of each column may be at a different height from the previous one. We cannot simply reposition point to the same line each time, but moved over to the right---or perhaps we can@dots{} We are planning to make the columns of the bar graph out of asterisks. The number of asterisks in the column is the number specified by the current element of the @code{numbers-list}. We need to construct a list of asterisks of the right length for each call to @code{insert-rectangle}. If this list consists solely of the requisite number of asterisks, then we will have to position point the right number of lines above the base for the graph to print correctly. This could be difficult. Alternatively, if we can figure out some way to pass @code{insert-rectangle} a list of the same length each time, then we can place point on the same line each time, but move it over one column to the right for each new column. If we do this, however, some of the entries in the list passed to @code{insert-rectangle} must be blanks rather than asterisks. For example, if the maximum height of the graph is 5, but the height of the column is 3, then @code{insert-rectangle} requires an argument that looks like this: @smallexample (" " " " "*" "*" "*") @end smallexample This last proposal is not so difficult, so long as we can determine the column height. There are two ways for us to specify the column height: we can arbitrarily state what it will be, which would work fine for graphs of that height; or we can search through the list of numbers and use the maximum height of the list as the maximum height of the graph. If the latter operation were difficult, then the former procedure would be easiest, but there is a function built into Emacs that determines the maximum of its arguments. We can use that function. The function is called @code{max} and it returns the largest of all its arguments, which must be numbers. Thus, for example, @smallexample (max 3 4 6 5 7 3) @end smallexample @noindent returns 7. (A corresponding function called @code{min} returns the smallest of all its arguments.) @findex max @findex min However, we cannot simply call @code{max} on the @code{numbers-list}; the @code{max} function expects numbers as its argument, not a list of numbers. Thus, the following expression, @smallexample (max '(3 4 6 5 7 3)) @end smallexample @need 800 @noindent produces the following error message; @smallexample Wrong type of argument: number-or-marker-p, (3 4 6 5 7 3) @end smallexample @findex apply We need a function that passes a list of arguments to a function. This function is @code{apply}. This function applies its first argument (a function) to its remaining arguments, the last of which may be a list. @need 1250 For example, @smallexample (apply 'max 3 4 7 3 '(4 8 5)) @end smallexample @noindent returns 8. (Incidentally, I don't know how you would learn of this function without a book such as this. It is possible to discover other functions, like @code{search-forward} or @code{insert-rectangle}, by guessing at a part of their names and then using @code{apropos}. Even though its base in metaphor is clear---apply its first argument to the rest---I doubt a novice would come up with that particular word when using @code{apropos} or other aid. Of course, I could be wrong; after all, the function was first named by someone who had to invent it.) The second and subsequent arguments to @code{apply} are optional, so we can use @code{apply} to call a function and pass the elements of a list to it, like this, which also returns 8: @smallexample (apply 'max '(4 8 5)) @end smallexample This latter way is how we will use @code{apply}. The @code{recursive-lengths-list-many-files} function returns a numbers' list to which we can apply @code{max} (we could also apply @code{max} to the sorted numbers' list; it does not matter whether the list is sorted or not.) @need 800 Hence, the operation for finding the maximum height of the graph is this: @smallexample (setq max-graph-height (apply 'max numbers-list)) @end smallexample Now we can return to the question of how to create a list of strings for a column of the graph. Told the maximum height of the graph and the number of asterisks that should appear in the column, the function should return a list of strings for the @code{insert-rectangle} command to insert. Each column is made up of asterisks or blanks. Since the function is passed the value of the height of the column and the number of asterisks in the column, the number of blanks can be found by subtracting the number of asterisks from the height of the column. Given the number of blanks and the number of asterisks, two @code{while} loops can be used to construct the list: @smallexample @group ;;; @r{First version.} (defun column-of-graph (max-graph-height actual-height) "Return list of strings that is one column of a graph." (let ((insert-list nil) (number-of-top-blanks (- max-graph-height actual-height))) @end group @group ;; @r{Fill in asterisks.} (while (> actual-height 0) (setq insert-list (cons "*" insert-list)) (setq actual-height (1- actual-height))) @end group @group ;; @r{Fill in blanks.} (while (> number-of-top-blanks 0) (setq insert-list (cons " " insert-list)) (setq number-of-top-blanks (1- number-of-top-blanks))) @end group @group ;; @r{Return whole list.} insert-list)) @end group @end smallexample If you install this function and then evaluate the following expression you will see that it returns the list as desired: @smallexample (column-of-graph 5 3) @end smallexample @need 800 @noindent returns @smallexample (" " " " "*" "*" "*") @end smallexample As written, @code{column-of-graph} contains a major flaw: the symbols used for the blank and for the marked entries in the column are hard-coded as a space and asterisk. This is fine for a prototype, but you, or another user, may wish to use other symbols. For example, in testing the graph function, you may want to use a period in place of the space, to make sure the point is being repositioned properly each time the @code{insert-rectangle} function is called; or you might want to substitute a @samp{+} sign or other symbol for the asterisk. You might even want to make a graph-column that is more than one display column wide. The program should be more flexible. The way to do that is to replace the blank and the asterisk with two variables that we can call @code{graph-blank} and @code{graph-symbol} and define those variables separately. Also, the documentation is not well written. These considerations lead us to the second version of the function: @smallexample @group (defvar graph-symbol "*" "String used as symbol in graph, usually an asterisk.") @end group @group (defvar graph-blank " " "String used as blank in graph, usually a blank space. graph-blank must be the same number of columns wide as graph-symbol.") @end group @end smallexample @noindent (For an explanation of @code{defvar}, see @ref{defvar, , Initializing a Variable with @code{defvar}}.) @smallexample @group ;;; @r{Second version.} (defun column-of-graph (max-graph-height actual-height) "Return MAX-GRAPH-HEIGHT strings; ACTUAL-HEIGHT are graph-symbols. @end group @group The graph-symbols are contiguous entries at the end of the list. The list will be inserted as one column of a graph. The strings are either graph-blank or graph-symbol." @end group @group (let ((insert-list nil) (number-of-top-blanks (- max-graph-height actual-height))) @end group @group ;; @r{Fill in @code{graph-symbols}.} (while (> actual-height 0) (setq insert-list (cons graph-symbol insert-list)) (setq actual-height (1- actual-height))) @end group @group ;; @r{Fill in @code{graph-blanks}.} (while (> number-of-top-blanks 0) (setq insert-list (cons graph-blank insert-list)) (setq number-of-top-blanks (1- number-of-top-blanks))) ;; @r{Return whole list.} insert-list)) @end group @end smallexample If we wished, we could rewrite @code{column-of-graph} a third time to provide optionally for a line graph as well as for a bar graph. This would not be hard to do. One way to think of a line graph is that it is no more than a bar graph in which the part of each bar that is below the top is blank. To construct a column for a line graph, the function first constructs a list of blanks that is one shorter than the value, then it uses @code{cons} to attach a graph symbol to the list; then it uses @code{cons} again to attach the top blanks to the list. It is easy to see how to write such a function, but since we don't need it, we will not do it. But the job could be done, and if it were done, it would be done with @code{column-of-graph}. Even more important, it is worth noting that few changes would have to be made anywhere else. The enhancement, if we ever wish to make it, is simple. Now, finally, we come to our first actual graph printing function. This prints the body of a graph, not the labels for the vertical and horizontal axes, so we can call this @code{graph-body-print}. @node graph-body-print @section The @code{graph-body-print} Function @findex graph-body-print After our preparation in the preceding section, the @code{graph-body-print} function is straightforward. The function will print column after column of asterisks and blanks, using the elements of a numbers' list to specify the number of asterisks in each column. This is a repetitive act, which means we can use a decrementing @code{while} loop or recursive function for the job. In this section, we will write the definition using a @code{while} loop. The @code{column-of-graph} function requires the height of the graph as an argument, so we should determine and record that as a local variable. This leads us to the following template for the @code{while} loop version of this function: @smallexample @group (defun graph-body-print (numbers-list) "@var{documentation}@dots{}" (let ((height @dots{} @dots{})) @end group @group (while numbers-list @var{insert-columns-and-reposition-point} (setq numbers-list (cdr numbers-list))))) @end group @end smallexample @noindent We need to fill in the slots of the template. Clearly, we can use the @code{(apply 'max numbers-list)} expression to determine the height of the graph. The @code{while} loop will cycle through the @code{numbers-list} one element at a time. As it is shortened by the @code{(setq numbers-list (cdr numbers-list))} expression, the @sc{car} of each instance of the list is the value of the argument for @code{column-of-graph}. At each cycle of the @code{while} loop, the @code{insert-rectangle} function inserts the list returned by @code{column-of-graph}. Since the @code{insert-rectangle} function moves point to the lower right of the inserted rectangle, we need to save the location of point at the time the rectangle is inserted, move back to that position after the rectangle is inserted, and then move horizontally to the next place from which @code{insert-rectangle} is called. If the inserted columns are one character wide, as they will be if single blanks and asterisks are used, the repositioning command is simply @code{(forward-char 1)}; however, the width of a column may be greater than one. This means that the repositioning command should be written @code{(forward-char symbol-width)}. The @code{symbol-width} itself is the length of a @code{graph-blank} and can be found using the expression @code{(length graph-blank)}. The best place to bind the @code{symbol-width} variable to the value of the width of graph column is in the varlist of the @code{let} expression. @need 1250 These considerations lead to the following function definition: @smallexample @group (defun graph-body-print (numbers-list) "Print a bar graph of the NUMBERS-LIST. The numbers-list consists of the Y-axis values." (let ((height (apply 'max numbers-list)) (symbol-width (length graph-blank)) from-position) @end group @group (while numbers-list (setq from-position (point)) (insert-rectangle (column-of-graph height (car numbers-list))) (goto-char from-position) (forward-char symbol-width) @end group @group ;; @r{Draw graph column by column.} (sit-for 0) (setq numbers-list (cdr numbers-list))) @end group @group ;; @r{Place point for X axis labels.} (forward-line height) (insert "\n") )) @end group @end smallexample @noindent The one unexpected expression in this function is the @w{@code{(sit-for 0)}} expression in the @code{while} loop. This expression makes the graph printing operation more interesting to watch than it would be otherwise. The expression causes Emacs to @dfn{sit} or do nothing for a zero length of time and then redraw the screen. Placed here, it causes Emacs to redraw the screen column by column. Without it, Emacs would not redraw the screen until the function exits. We can test @code{graph-body-print} with a short list of numbers. @enumerate @item Install @code{graph-symbol}, @code{graph-blank}, @code{column-of-graph}, which are in @iftex @ref{Readying a Graph, , Readying a Graph}, @end iftex @ifnottex @ref{Columns of a graph}, @end ifnottex and @code{graph-body-print}. @need 800 @item Copy the following expression: @smallexample (graph-body-print '(1 2 3 4 6 4 3 5 7 6 5 2 3)) @end smallexample @item Switch to the @file{*scratch*} buffer and place the cursor where you want the graph to start. @item Type @kbd{M-:} (@code{eval-expression}). @item Yank the @code{graph-body-print} expression into the minibuffer with @kbd{C-y} (@code{yank)}. @item Press @key{RET} to evaluate the @code{graph-body-print} expression. @end enumerate @need 800 Emacs will print a graph like this: @smallexample @group * * ** * **** *** **** ********* * ************ ************* @end group @end smallexample @node recursive-graph-body-print @section The @code{recursive-graph-body-print} Function @findex recursive-graph-body-print The @code{graph-body-print} function may also be written recursively. The recursive solution is divided into two parts: an outside wrapper that uses a @code{let} expression to determine the values of several variables that need only be found once, such as the maximum height of the graph, and an inside function that is called recursively to print the graph. @need 1250 The wrapper is uncomplicated: @smallexample @group (defun recursive-graph-body-print (numbers-list) "Print a bar graph of the NUMBERS-LIST. The numbers-list consists of the Y-axis values." (let ((height (apply 'max numbers-list)) (symbol-width (length graph-blank)) from-position) (recursive-graph-body-print-internal numbers-list height symbol-width))) @end group @end smallexample The recursive function is a little more difficult. It has four parts: the do-again-test, the printing code, the recursive call, and the next-step-expression. The do-again-test is a @code{when} expression that determines whether the @code{numbers-list} contains any remaining elements; if it does, the function prints one column of the graph using the printing code and calls itself again. The function calls itself again according to the value produced by the next-step-expression which causes the call to act on a shorter version of the @code{numbers-list}. @smallexample @group (defun recursive-graph-body-print-internal (numbers-list height symbol-width) "Print a bar graph. Used within recursive-graph-body-print function." @end group @group (when numbers-list (setq from-position (point)) (insert-rectangle (column-of-graph height (car numbers-list))) @end group @group (goto-char from-position) (forward-char symbol-width) (sit-for 0) ; @r{Draw graph column by column.} (recursive-graph-body-print-internal (cdr numbers-list) height symbol-width))) @end group @end smallexample @need 1250 After installation, this expression can be tested; here is a sample: @smallexample (recursive-graph-body-print '(3 2 5 6 7 5 3 4 6 4 3 2 1)) @end smallexample @need 800 Here is what @code{recursive-graph-body-print} produces: @smallexample @group * ** * **** * **** *** * ********* ************ ************* @end group @end smallexample Either of these two functions, @code{graph-body-print} or @code{recursive-graph-body-print}, create the body of a graph. @node Printed Axes @section Need for Printed Axes A graph needs printed axes, so you can orient yourself. For a do-once project, it may be reasonable to draw the axes by hand using Emacs's Picture mode; but a graph drawing function may be used more than once. For this reason, I have written enhancements to the basic @code{print-graph-body} function that automatically print labels for the horizontal and vertical axes. Since the label printing functions do not contain much new material, I have placed their description in an appendix. @xref{Full Graph, , A Graph with Labeled Axes}. @node Line Graph Exercise @section Exercise Write a line graph version of the graph printing functions. @node Emacs Initialization @chapter Your @file{.emacs} File @cindex @file{.emacs} file @cindex Customizing your @file{.emacs} file @cindex Initialization file ``You don't have to like Emacs to like it''---this seemingly paradoxical statement is the secret of GNU Emacs. The plain, out-of-the-box Emacs is a generic tool. Most people who use it customize it to suit themselves. GNU Emacs is mostly written in Emacs Lisp; this means that by writing expressions in Emacs Lisp you can change or extend Emacs. @menu * Default Configuration:: * Site-wide Init:: You can write site-wide init files. * defcustom:: Emacs will write code for you. * Beginning init File:: How to write a @file{.emacs} init file. * Text and Auto-fill:: Automatically wrap lines. * Mail Aliases:: Use abbreviations for email addresses. * Indent Tabs Mode:: Don't use tabs with @TeX{} * Keybindings:: Create some personal keybindings. * Keymaps:: More about key binding. * Loading Files:: Load (i.e., evaluate) files automatically. * Autoload:: Make functions available. * Simple Extension:: Define a function; bind it to a key. * X11 Colors:: Colors in X. * Miscellaneous:: * Mode Line:: How to customize your mode line. @end menu @ifnottex @node Default Configuration @unnumberedsec Emacs's Default Configuration @end ifnottex There are those who appreciate Emacs's default configuration. After all, Emacs starts you in C mode when you edit a C file, starts you in Fortran mode when you edit a Fortran file, and starts you in Fundamental mode when you edit an unadorned file. This all makes sense, if you do not know who is going to use Emacs. Who knows what a person hopes to do with an unadorned file? Fundamental mode is the right default for such a file, just as C mode is the right default for editing C code. (Enough programming languages have syntaxes that enable them to share or nearly share features, so C mode is now provided by CC mode, the C Collection.) But when you do know who is going to use Emacs---you, yourself---then it makes sense to customize Emacs. For example, I seldom want Fundamental mode when I edit an otherwise undistinguished file; I want Text mode. This is why I customize Emacs: so it suits me. You can customize and extend Emacs by writing or adapting a @file{~/.emacs} file. This is your personal initialization file; its contents, written in Emacs Lisp, tell Emacs what to do.@footnote{You may also add @file{.el} to @file{~/.emacs} and call it a @file{~/.emacs.el} file. In the past, you were forbidden to type the extra keystrokes that the name @file{~/.emacs.el} requires, but now you may. The new format is consistent with the Emacs Lisp file naming conventions; the old format saves typing.} A @file{~/.emacs} file contains Emacs Lisp code. You can write this code yourself; or you can use Emacs's @code{customize} feature to write the code for you. You can combine your own expressions and auto-written Customize expressions in your @file{.emacs} file. (I myself prefer to write my own expressions, except for those, particularly fonts, that I find easier to manipulate using the @code{customize} command. I combine the two methods.) Most of this chapter is about writing expressions yourself. It describes a simple @file{.emacs} file; for more information, see @ref{Init File, , The Init File, emacs, The GNU Emacs Manual}, and @ref{Init File, , The Init File, elisp, The GNU Emacs Lisp Reference Manual}. @node Site-wide Init @section Site-wide Initialization Files @cindex @file{default.el} init file @cindex @file{site-init.el} init file @cindex @file{site-load.el} init file In addition to your personal initialization file, Emacs automatically loads various site-wide initialization files, if they exist. These have the same form as your @file{.emacs} file, but are loaded by everyone. Two site-wide initialization files, @file{site-load.el} and @file{site-init.el}, are loaded into Emacs and then dumped if a dumped version of Emacs is created, as is most common. (Dumped copies of Emacs load more quickly. However, once a file is loaded and dumped, a change to it does not lead to a change in Emacs unless you load it yourself or re-dump Emacs. @xref{Building Emacs, , Building Emacs, elisp, The GNU Emacs Lisp Reference Manual}, and the @file{INSTALL} file.) Three other site-wide initialization files are loaded automatically each time you start Emacs, if they exist. These are @file{site-start.el}, which is loaded @emph{before} your @file{.emacs} file, and @file{default.el}, and the terminal type file, which are both loaded @emph{after} your @file{.emacs} file. Settings and definitions in your @file{.emacs} file will overwrite conflicting settings and definitions in a @file{site-start.el} file, if it exists; but the settings and definitions in a @file{default.el} or terminal type file will overwrite those in your @file{.emacs} file. (You can prevent interference from a terminal type file by setting @code{term-file-prefix} to @code{nil}. @xref{Simple Extension, , A Simple Extension}.) @c Rewritten to avoid overfull hbox. The @file{INSTALL} file that comes in the distribution contains descriptions of the @file{site-init.el} and @file{site-load.el} files. The @file{loadup.el}, @file{startup.el}, and @file{loaddefs.el} files control loading. These files are in the @file{lisp} directory of the Emacs distribution and are worth perusing. The @file{loaddefs.el} file contains a good many suggestions as to what to put into your own @file{.emacs} file, or into a site-wide initialization file. @node defcustom @section Specifying Variables using @code{defcustom} @findex defcustom You can specify variables using @code{defcustom} so that you and others can then use Emacs's @code{customize} feature to set their values. (You cannot use @code{customize} to write function definitions; but you can write @code{defuns} in your @file{.emacs} file. Indeed, you can write any Lisp expression in your @file{.emacs} file.) The @code{customize} feature depends on the @code{defcustom} macro. Although you can use @code{defvar} or @code{setq} for variables that users set, the @code{defcustom} macro is designed for the job. You can use your knowledge of @code{defvar} for writing the first three arguments for @code{defcustom}. The first argument to @code{defcustom} is the name of the variable. The second argument is the variable's initial value, if any; and this value is set only if the value has not already been set. The third argument is the documentation. The fourth and subsequent arguments to @code{defcustom} specify types and options; these are not featured in @code{defvar}. (These arguments are optional.) Each of these arguments consists of a keyword followed by a value. Each keyword starts with the colon character @samp{:}. @need 1250 For example, the customizable user option variable @code{text-mode-hook} looks like this: @smallexample @group (defcustom text-mode-hook nil "Normal hook run when entering Text mode and many related modes." :type 'hook :options '(turn-on-auto-fill flyspell-mode) :group 'wp) @end group @end smallexample @noindent The name of the variable is @code{text-mode-hook}; it has no default value; and its documentation string tells you what it does. The @code{:type} keyword tells Emacs the kind of data to which @code{text-mode-hook} should be set and how to display the value in a Customization buffer. The @code{:options} keyword specifies a suggested list of values for the variable. Usually, @code{:options} applies to a hook. The list is only a suggestion; it is not exclusive; a person who sets the variable may set it to other values; the list shown following the @code{:options} keyword is intended to offer convenient choices to a user. Finally, the @code{:group} keyword tells the Emacs Customization command in which group the variable is located. This tells where to find it. The @code{defcustom} macro recognizes more than a dozen keywords. For more information, see @ref{Customization, , Writing Customization Definitions, elisp, The GNU Emacs Lisp Reference Manual}. Consider @code{text-mode-hook} as an example. There are two ways to customize this variable. You can use the customization command or write the appropriate expressions yourself. @need 800 Using the customization command, you can type: @smallexample M-x customize @end smallexample @noindent and find that the group for editing files of text is called ``Text''. Enter that group. Text Mode Hook is the first member. You can click on its various options, such as @code{turn-on-auto-fill}, to set the values. After you click on the button to @smallexample Save for Future Sessions @end smallexample @noindent Emacs will write an expression into your @file{.emacs} file. It will look like this: @smallexample @group (custom-set-variables ;; custom-set-variables was added by Custom. ;; If you edit it by hand, you could mess it up, so be careful. ;; Your init file should contain only one such instance. ;; If there is more than one, they won't work right. '(text-mode-hook '(turn-on-auto-fill text-mode-hook-identify))) @end group @end smallexample @noindent (The @code{text-mode-hook-identify} function tells @code{toggle-text-mode-auto-fill} which buffers are in Text mode. It comes on automatically.) The @code{custom-set-variables} function works somewhat differently than a @code{setq}. While I have never learned the differences, I modify the @code{custom-set-variables} expressions in my @file{.emacs} file by hand: I make the changes in what appears to me to be a reasonable manner and have not had any problems. Others prefer to use the Customization command and let Emacs do the work for them. Another @code{custom-set-@dots{}} function is @code{custom-set-faces}. This function sets the various font faces. Over time, I have set a considerable number of faces. Some of the time, I re-set them using @code{customize}; other times, I simply edit the @code{custom-set-faces} expression in my @file{.emacs} file itself. The second way to customize your @code{text-mode-hook} is to set it yourself in your @file{.emacs} file using code that has nothing to do with the @code{custom-set-@dots{}} functions. @need 800 When you do this, and later use @code{customize}, you will see a message that says @smallexample CHANGED outside Customize; operating on it here may be unreliable. @end smallexample @need 800 This message is only a warning. If you click on the button to @smallexample Save for Future Sessions @end smallexample @noindent Emacs will write a @code{custom-set-@dots{}} expression near the end of your @file{.emacs} file that will be evaluated after your hand-written expression. It will, therefore, overrule your hand-written expression. No harm will be done. When you do this, however, be careful to remember which expression is active; if you forget, you may confuse yourself. So long as you remember where the values are set, you will have no trouble. In any event, the values are always set in your initialization file, which is usually called @file{.emacs}. I myself use @code{customize} for hardly anything. Mostly, I write expressions myself. @findex defsubst @findex defconst Incidentally, to be more complete concerning defines: @code{defsubst} defines an inline function. The syntax is just like that of @code{defun}. @code{defconst} defines a symbol as a constant. The intent is that neither programs nor users should ever change a value set by @code{defconst}. (You can change it; the value set is a variable; but please do not.) @node Beginning init File @section Beginning a @file{.emacs} File @cindex @file{.emacs} file, beginning of When you start Emacs, it loads your @file{.emacs} file unless you tell it not to by specifying @samp{-q} on the command line. (The @code{emacs -q} command gives you a plain, out-of-the-box Emacs.) A @file{.emacs} file contains Lisp expressions. Often, these are no more than expressions to set values; sometimes they are function definitions. @xref{Init File, , The Init File @file{~/.emacs}, emacs, The GNU Emacs Manual}, for a short description of initialization files. This chapter goes over some of the same ground, but is a walk among extracts from a complete, long-used @file{.emacs} file---my own. The first part of the file consists of comments: reminders to myself. By now, of course, I remember these things, but when I started, I did not. @need 1200 @smallexample @group ;;;; Bob's .emacs file ; Robert J. Chassell ; 26 September 1985 @end group @end smallexample @noindent Look at that date! I started this file a long time ago. I have been adding to it ever since. @smallexample @group ; Each section in this file is introduced by a ; line beginning with four semicolons; and each ; entry is introduced by a line beginning with ; three semicolons. @end group @end smallexample @noindent This describes the usual conventions for comments in Emacs Lisp. Everything on a line that follows a semicolon is a comment. Two, three, and four semicolons are used as subsection and section markers. (@xref{Comments, ,, elisp, The GNU Emacs Lisp Reference Manual}, for more about comments.) @smallexample @group ;;;; The Help Key ; Control-h is the help key; ; after typing control-h, type a letter to ; indicate the subject about which you want help. ; For an explanation of the help facility, ; type control-h two times in a row. @end group @end smallexample @noindent Just remember: type @kbd{C-h} two times for help. @smallexample @group ; To find out about any mode, type control-h m ; while in that mode. For example, to find out ; about mail mode, enter mail mode and then type ; control-h m. @end group @end smallexample @noindent ``Mode help'', as I call this, is very helpful. Usually, it tells you all you need to know. Of course, you don't need to include comments like these in your @file{.emacs} file. I included them in mine because I kept forgetting about Mode help or the conventions for comments---but I was able to remember to look here to remind myself. @node Text and Auto-fill @section Text and Auto Fill Mode Now we come to the part that turns on Text mode and Auto Fill mode. @smallexample @group ;;; Text mode and Auto Fill mode ;; The next two lines put Emacs into Text mode ;; and Auto Fill mode, and are for writers who ;; want to start writing prose rather than code. (setq-default major-mode 'text-mode) (add-hook 'text-mode-hook 'turn-on-auto-fill) @end group @end smallexample Here is the first part of this @file{.emacs} file that does something besides remind a forgetful human! The first of the two lines in parentheses tells Emacs to turn on Text mode when you find a file, @emph{unless} that file should go into some other mode, such as C mode. @cindex Per-buffer, local variables list @cindex Local variables list, per-buffer, @cindex Automatic mode selection @cindex Mode selection, automatic When Emacs reads a file, it looks at the extension to the file name, if any. (The extension is the part that comes after a @samp{.}.) If the file ends with a @samp{.c} or @samp{.h} extension then Emacs turns on C mode. Also, Emacs looks at first nonblank line of the file; if the line says @w{@samp{-*- C -*-}}, Emacs turns on C mode. Emacs possesses a list of extensions and specifications that it uses automatically. In addition, Emacs looks near the last page for a per-buffer, local variables list, if any. @ifinfo @xref{Choosing Modes, , How Major Modes are Chosen, emacs, The GNU Emacs Manual}. @xref{File Variables, , Local Variables in Files, emacs, The GNU Emacs Manual}. @end ifinfo @iftex See sections ``How Major Modes are Chosen'' and ``Local Variables in Files'' in @cite{The GNU Emacs Manual}. @end iftex Now, back to the @file{.emacs} file. @need 800 Here is the line again; how does it work? @cindex Text Mode turned on @smallexample (setq major-mode 'text-mode) @end smallexample @noindent This line is a short, but complete Emacs Lisp expression. We are already familiar with @code{setq}. It sets the following variable, @code{major-mode}, to the subsequent value, which is @code{text-mode}. The single-quote before @code{text-mode} tells Emacs to deal directly with the @code{text-mode} symbol, not with whatever it might stand for. @xref{set & setq, , Setting the Value of a Variable}, for a reminder of how @code{setq} works. The main point is that there is no difference between the procedure you use to set a value in your @file{.emacs} file and the procedure you use anywhere else in Emacs. @need 800 Here is the next line: @cindex Auto Fill mode turned on @findex add-hook @smallexample (add-hook 'text-mode-hook 'turn-on-auto-fill) @end smallexample @noindent In this line, the @code{add-hook} command adds @code{turn-on-auto-fill} to the variable. @code{turn-on-auto-fill} is the name of a program, that, you guessed it!, turns on Auto Fill mode. Every time Emacs turns on Text mode, Emacs runs the commands hooked onto Text mode. So every time Emacs turns on Text mode, Emacs also turns on Auto Fill mode. In brief, the first line causes Emacs to enter Text mode when you edit a file, unless the file name extension, a first non-blank line, or local variables to tell Emacs otherwise. Text mode among other actions, sets the syntax table to work conveniently for writers. In Text mode, Emacs considers an apostrophe as part of a word like a letter; but Emacs does not consider a period or a space as part of a word. Thus, @kbd{M-f} moves you over @samp{it's}. On the other hand, in C mode, @kbd{M-f} stops just after the @samp{t} of @samp{it's}. The second line causes Emacs to turn on Auto Fill mode when it turns on Text mode. In Auto Fill mode, Emacs automatically breaks a line that is too wide and brings the excessively wide part of the line down to the next line. Emacs breaks lines between words, not within them. When Auto Fill mode is turned off, lines continue to the right as you type them. Depending on how you set the value of @code{truncate-lines}, the words you type either disappear off the right side of the screen, or else are shown, in a rather ugly and unreadable manner, as a continuation line on the screen. @need 1250 In addition, in this part of my @file{.emacs} file, I tell the Emacs fill commands to insert two spaces after a colon: @smallexample (setq colon-double-space t) @end smallexample @node Mail Aliases @section Mail Aliases Here is a @code{setq} that turns on mail aliases, along with more reminders. @smallexample @group ;;; Message mode ; To enter message mode, type 'C-x m' ; To enter RMAIL (for reading mail), ; type 'M-x rmail' (setq mail-aliases t) @end group @end smallexample @cindex Mail aliases @noindent This @code{setq} sets the value of the variable @code{mail-aliases} to @code{t}. Since @code{t} means true, the line says, in effect, ``Yes, use mail aliases.'' Mail aliases are convenient short names for long email addresses or for lists of email addresses. The file where you keep your aliases is @file{~/.mailrc}. You write an alias like this: @smallexample alias geo george@@foobar.wiz.edu @end smallexample @noindent When you write a message to George, address it to @samp{geo}; the mailer will automatically expand @samp{geo} to the full address. @node Indent Tabs Mode @section Indent Tabs Mode @cindex Tabs, preventing @findex indent-tabs-mode By default, Emacs inserts tabs in place of multiple spaces when it formats a region. (For example, you might indent many lines of text all at once with the @code{indent-region} command.) Tabs look fine on a terminal or with ordinary printing, but they produce badly indented output when you use @TeX{} or Texinfo since @TeX{} ignores tabs. @need 1250 The following turns off Indent Tabs mode: @smallexample @group ;;; Prevent Extraneous Tabs (setq-default indent-tabs-mode nil) @end group @end smallexample Note that this line uses @code{setq-default} rather than the @code{setq} that we have seen before; @code{setq-default} sets values only in buffers that do not have their own local values for the variable. @ifinfo @xref{Just Spaces, , Tabs vs.@: Spaces, emacs, The GNU Emacs Manual}. @xref{File Variables, , Local Variables in Files, emacs, The GNU Emacs Manual}. @end ifinfo @iftex See sections ``Tabs vs.@: Spaces'' and ``Local Variables in Files'' in @cite{The GNU Emacs Manual}. @end iftex @need 1700 @node Keybindings @section Some Keybindings Now for some personal keybindings: @smallexample @group ;;; Compare windows (global-set-key "\C-cw" 'compare-windows) @end group @end smallexample @findex compare-windows @code{compare-windows} is a nifty command that compares the text in your current window with text in the next window. It makes the comparison by starting at point in each window, moving over text in each window as far as they match. I use this command all the time. This also shows how to set a key globally, for all modes. @cindex Setting a key globally @cindex Global set key @cindex Key setting globally @findex global-set-key The command is @code{global-set-key}. It is followed by the keybinding. In a @file{.emacs} file, the keybinding is written as shown: @code{\C-c} stands for Control-C, which means to press the control key and the @kbd{c} key at the same time. The @code{w} means to press the @kbd{w} key. The keybinding is surrounded by double quotation marks. In documentation, you would write this as @w{@kbd{C-c w}}. (If you were binding a @key{META} key, such as @kbd{M-c}, rather than a @key{CTRL} key, you would write @w{@code{\M-c}} in your @file{.emacs} file. @xref{Init Rebinding, , Rebinding Keys in Your Init File, emacs, The GNU Emacs Manual}, for details.) The command invoked by the keys is @code{compare-windows}. Note that @code{compare-windows} is preceded by a single-quote; otherwise, Emacs would first try to evaluate the symbol to determine its value. These three things, the double quotation marks, the backslash before the @samp{C}, and the single-quote are necessary parts of keybinding that I tend to forget. Fortunately, I have come to remember that I should look at my existing @file{.emacs} file, and adapt what is there. As for the keybinding itself: @kbd{C-c w}. This combines the prefix key, @kbd{C-c}, with a single character, in this case, @kbd{w}. This set of keys, @kbd{C-c} followed by a single character, is strictly reserved for individuals' own use. (I call these @dfn{own} keys, since these are for my own use.) You should always be able to create such a keybinding for your own use without stomping on someone else's keybinding. If you ever write an extension to Emacs, please avoid taking any of these keys for public use. Create a key like @kbd{C-c C-w} instead. Otherwise, we will run out of own keys. @need 1250 Here is another keybinding, with a comment: @smallexample @group ;;; Keybinding for 'occur' ; I use occur a lot, so let's bind it to a key: (global-set-key "\C-co" 'occur) @end group @end smallexample @findex occur The @code{occur} command shows all the lines in the current buffer that contain a match for a regular expression. When the region is active, @code{occur} restricts matches to such region. Otherwise it uses the entire buffer. Matching lines are shown in a buffer called @file{*Occur*}. That buffer serves as a menu to jump to occurrences. @findex global-unset-key @cindex Unbinding key @cindex Key unbinding @need 1250 Here is how to unbind a key, so it does not work: @smallexample @group ;;; Unbind 'C-x f' (global-unset-key "\C-xf") @end group @end smallexample There is a reason for this unbinding: I found I inadvertently typed @w{@kbd{C-x f}} when I meant to type @kbd{C-x C-f}. Rather than find a file, as I intended, I accidentally set the width for filled text, almost always to a width I did not want. Since I hardly ever reset my default width, I simply unbound the key. @findex list-buffers@r{, rebound} @findex buffer-menu@r{, bound to key} @need 1250 The following rebinds an existing key: @smallexample @group ;;; Rebind 'C-x C-b' for 'buffer-menu' (global-set-key "\C-x\C-b" 'buffer-menu) @end group @end smallexample By default, @kbd{C-x C-b} runs the @code{list-buffers} command. This command lists your buffers in @emph{another} window. Since I almost always want to do something in that window, I prefer the @code{buffer-menu} command, which not only lists the buffers, but moves point into that window. @node Keymaps @section Keymaps @cindex Keymaps @cindex Rebinding keys Emacs uses @dfn{keymaps} to record which keys call which commands. When you use @code{global-set-key} to set the keybinding for a single command in all parts of Emacs, you are specifying the keybinding in @code{current-global-map}. Specific modes, such as C mode or Text mode, have their own keymaps; the mode-specific keymaps override the global map that is shared by all buffers. The @code{global-set-key} function binds, or rebinds, the global keymap. For example, the following binds the key @kbd{C-x C-b} to the function @code{buffer-menu}: @smallexample (global-set-key "\C-x\C-b" 'buffer-menu) @end smallexample Mode-specific keymaps are bound using the @code{define-key} function, which takes a specific keymap as an argument, as well as the key and the command. For example, my @file{.emacs} file contains the following expression to bind the @code{texinfo-insert-@@group} command to @kbd{C-c C-c g}: @smallexample @group (define-key texinfo-mode-map "\C-c\C-cg" 'texinfo-insert-@@group) @end group @end smallexample @noindent The @code{texinfo-insert-@@group} function itself is a little extension to Texinfo mode that inserts @samp{@@group} into a Texinfo file. I use this command all the time and prefer to type the three strokes @kbd{C-c C-c g} rather than the six strokes @kbd{@@ g r o u p}. (@samp{@@group} and its matching @samp{@@end group} are commands that keep all enclosed text together on one page; many multi-line examples in this book are surrounded by @samp{@@group @dots{} @@end group}.) @need 1250 Here is the @code{texinfo-insert-@@group} function definition: @smallexample @group (defun texinfo-insert-@@group () "Insert the string @@group in a Texinfo buffer." (interactive) (beginning-of-line) (insert "@@group\n")) @end group @end smallexample (Of course, I could have used Abbrev mode to save typing, rather than write a function to insert a word; but I prefer key strokes consistent with other Texinfo mode key bindings.) You will see numerous @code{define-key} expressions in @file{loaddefs.el} as well as in the various mode libraries, such as @file{cc-mode.el} and @file{lisp-mode.el}. @xref{Key Bindings, , Customizing Key Bindings, emacs, The GNU Emacs Manual}, and @ref{Keymaps, , Keymaps, elisp, The GNU Emacs Lisp Reference Manual}, for more information about keymaps. @node Loading Files @section Loading Files @cindex Loading files @c findex load Many people in the GNU Emacs community have written extensions to Emacs. As time goes by, these extensions are often included in new releases. For example, the Calendar and Diary packages are now part of the standard GNU Emacs, as is Calc. You can use a @code{load} command to evaluate a complete file and thereby install all the functions and variables in the file into Emacs. For example: @c (auto-compression-mode t) @smallexample (load "~/emacs/slowsplit") @end smallexample This evaluates, i.e., loads, the @file{slowsplit.el} file or if it exists, the faster, byte compiled @file{slowsplit.elc} file from the @file{emacs} sub-directory of your home directory. The file contains the function @code{split-window-quietly}, which John Robinson wrote in 1989. The @code{split-window-quietly} function splits a window with the minimum of redisplay. I installed it in 1989 because it worked well with the slow 1200 baud terminals I was then using. Nowadays, I only occasionally come across such a slow connection, but I continue to use the function because I like the way it leaves the bottom half of a buffer in the lower of the new windows and the top half in the upper window. @need 1250 To replace the key binding for the default @code{split-window-vertically}, you must also unset that key and bind the keys to @code{split-window-quietly}, like this: @smallexample @group (global-unset-key "\C-x2") (global-set-key "\C-x2" 'split-window-quietly) @end group @end smallexample @vindex load-path If you load many extensions, as I do, then instead of specifying the exact location of the extension file, as shown above, you can specify that directory as part of Emacs's @code{load-path}. Then, when Emacs loads a file, it will search that directory as well as its default list of directories. (The default list is specified in @file{paths.h} when Emacs is built.) @need 1250 The following command adds your @file{~/emacs} directory to the existing load path: @smallexample @group ;;; Emacs Load Path (setq load-path (cons "~/emacs" load-path)) @end group @end smallexample Incidentally, @code{load-library} is an interactive interface to the @code{load} function. The complete function looks like this: @findex load-library @smallexample @group (defun load-library (library) "Load the Emacs Lisp library named LIBRARY. This is an interface to the function `load'. LIBRARY is searched for in `load-path', both with and without `load-suffixes' (as well as `load-file-rep-suffixes'). See Info node `(emacs)Lisp Libraries' for more details. See `load-file' for a different interface to `load'." (interactive (list (completing-read "Load library: " (apply-partially 'locate-file-completion-table load-path (get-load-suffixes))))) (load library)) @end group @end smallexample The name of the function, @code{load-library}, comes from the use of ``library'' as a conventional synonym for ``file''. The source for the @code{load-library} command is in the @file{files.el} library. Another interactive command that does a slightly different job is @code{load-file}. @xref{Lisp Libraries, , Libraries of Lisp Code for Emacs, emacs, The GNU Emacs Manual}, for information on the distinction between @code{load-library} and this command. @node Autoload @section Autoloading @findex autoload Instead of installing a function by loading the file that contains it, or by evaluating the function definition, you can make the function available but not actually install it until it is first called. This is called @dfn{autoloading}. When you execute an autoloaded function, Emacs automatically evaluates the file that contains the definition, and then calls the function. Emacs starts quicker with autoloaded functions, since their libraries are not loaded right away; but you need to wait a moment when you first use such a function, while its containing file is evaluated. Rarely used functions are frequently autoloaded. The @file{loaddefs.el} library contains thousands of autoloaded functions, from @code{5x5} to @code{zone}. Of course, you may come to use a rare function frequently. When you do, you should load that function's file with a @code{load} expression in your @file{.emacs} file. In my @file{.emacs} file, I load 14 libraries that contain functions that would otherwise be autoloaded. (Actually, it would have been better to include these files in my dumped Emacs, but I forgot. @xref{Building Emacs, , Building Emacs, elisp, The GNU Emacs Lisp Reference Manual}, and the @file{INSTALL} file for more about dumping.) You may also want to include autoloaded expressions in your @file{.emacs} file. @code{autoload} is a built-in function that takes up to five arguments, the final three of which are optional. The first argument is the name of the function to be autoloaded; the second is the name of the file to be loaded. The third argument is documentation for the function, and the fourth tells whether the function can be called interactively. The fifth argument tells what type of object---@code{autoload} can handle a keymap or macro as well as a function (the default is a function). @need 800 Here is a typical example: @smallexample @group (autoload 'html-helper-mode "html-helper-mode" "Edit HTML documents" t) @end group @end smallexample @noindent (@code{html-helper-mode} is an older alternative to @code{html-mode}, which is a standard part of the distribution.) @noindent This expression autoloads the @code{html-helper-mode} function. It takes it from the @file{html-helper-mode.el} file (or from the byte compiled version @file{html-helper-mode.elc}, if that exists.) The file must be located in a directory specified by @code{load-path}. The documentation says that this is a mode to help you edit documents written in the HyperText Markup Language. You can call this mode interactively by typing @kbd{M-x html-helper-mode}. (You need to duplicate the function's regular documentation in the autoload expression because the regular function is not yet loaded, so its documentation is not available.) @xref{Autoload, , Autoload, elisp, The GNU Emacs Lisp Reference Manual}, for more information. @node Simple Extension @section A Simple Extension: @code{line-to-top-of-window} @findex line-to-top-of-window @cindex Simple extension in @file{.emacs} file Here is a simple extension to Emacs that moves the line point is on to the top of the window. I use this all the time, to make text easier to read. You can put the following code into a separate file and then load it from your @file{.emacs} file, or you can include it within your @file{.emacs} file. @need 1250 Here is the definition: @smallexample @group ;;; Line to top of window; ;;; replace three keystroke sequence C-u 0 C-l (defun line-to-top-of-window () "Move the line point is on to top of window." (interactive) (recenter 0)) @end group @end smallexample @need 1250 Now for the keybinding. Nowadays, function keys as well as mouse button events and non-@sc{ascii} characters are written within square brackets, without quotation marks. (In Emacs version 18 and before, you had to write different function key bindings for each different make of terminal.) I bind @code{line-to-top-of-window} to my @key{F6} function key like this: @smallexample (global-set-key [f6] 'line-to-top-of-window) @end smallexample For more information, see @ref{Init Rebinding, , Rebinding Keys in Your Init File, emacs, The GNU Emacs Manual}. @cindex Conditional 'twixt two versions of Emacs @cindex Version of Emacs, choosing @cindex Emacs version, choosing If you run two versions of GNU Emacs, such as versions 22 and 23, and use one @file{.emacs} file, you can select which code to evaluate with the following conditional: @smallexample @group (cond ((= 22 emacs-major-version) ;; evaluate version 22 code ( @dots{} )) ((= 23 emacs-major-version) ;; evaluate version 23 code ( @dots{} ))) @end group @end smallexample For example, recent versions blink their cursors by default. I hate such blinking, as well as other features, so I placed the following in my @file{.emacs} file@footnote{When I start instances of Emacs that do not load my @file{.emacs} file or any site file, I also turn off blinking: @smallexample emacs -q --no-site-file -eval '(blink-cursor-mode nil)' @exdent Or nowadays, using an even more sophisticated set of options, emacs -Q -D @end smallexample }: @smallexample @group (when (>= emacs-major-version 21) (blink-cursor-mode 0) ;; Insert newline when you press 'C-n' (next-line) ;; at the end of the buffer (setq next-line-add-newlines t) @end group @group ;; Turn on image viewing (auto-image-file-mode t) @end group @group ;; Turn on menu bar (this bar has text) ;; (Use numeric argument to turn on) (menu-bar-mode 1) @end group @group ;; Turn off tool bar (this bar has icons) ;; (Use numeric argument to turn on) (tool-bar-mode nil) @end group @group ;; Turn off tooltip mode for tool bar ;; (This mode causes icon explanations to pop up) ;; (Use numeric argument to turn on) (tooltip-mode nil) ;; If tooltips turned on, make tips appear promptly (setq tooltip-delay 0.1) ; default is 0.7 second ) @end group @end smallexample @node X11 Colors @section X11 Colors You can specify colors when you use Emacs with the MIT X Windowing system. I dislike the default colors and specify my own. @need 1250 Here are the expressions in my @file{.emacs} file that set values: @smallexample @group ;; Set cursor color (set-cursor-color "white") ;; Set mouse color (set-mouse-color "white") ;; Set foreground and background (set-foreground-color "white") (set-background-color "darkblue") @end group @group ;;; Set highlighting colors for isearch and drag (set-face-foreground 'highlight "white") (set-face-background 'highlight "blue") @end group @group (set-face-foreground 'region "cyan") (set-face-background 'region "blue") @end group @group (set-face-foreground 'secondary-selection "skyblue") (set-face-background 'secondary-selection "darkblue") @end group @group ;; Set calendar highlighting colors (with-eval-after-load 'calendar (set-face-foreground 'diary "skyblue") (set-face-background 'holiday "slate blue") (set-face-foreground 'holiday "white")) @end group @end smallexample The various shades of blue soothe my eye and prevent me from seeing the screen flicker. Alternatively, I could have set my specifications in various X initialization files. For example, I could set the foreground, background, cursor, and pointer (i.e., mouse) colors in my @file{~/.Xresources} file like this: @smallexample @group Emacs*foreground: white Emacs*background: darkblue Emacs*cursorColor: white Emacs*pointerColor: white @end group @end smallexample In any event, since it is not part of Emacs, I set the root color of my X window in my @file{~/.xinitrc} file, like this@footnote{I also run more modern window managers, such as Enlightenment, Gnome, or KDE; in those cases, I often specify an image rather than a plain color.}: @smallexample xsetroot -solid Navy -fg white & @end smallexample @need 1700 @node Miscellaneous @section Miscellaneous Settings for a @file{.emacs} File @need 1250 Here are a few miscellaneous settings: @sp 1 @itemize @minus @item Set the shape and color of the mouse cursor: @smallexample @group ; Cursor shapes are defined in ; '/usr/include/X11/cursorfont.h'; ; for example, the 'target' cursor is number 128; ; the 'top_left_arrow' cursor is number 132. @end group @group (let ((mpointer (x-get-resource "*mpointer" "*emacs*mpointer"))) ;; If you have not set your mouse pointer ;; then set it, otherwise leave as is: (if (eq mpointer nil) (setq mpointer "132")) ; top_left_arrow @end group @group (setq x-pointer-shape (string-to-number mpointer)) (set-mouse-color "white")) @end group @end smallexample @item Or you can set the values of a variety of features in an alist, like this: @smallexample @group (setq-default default-frame-alist '((cursor-color . "white") (mouse-color . "white") (foreground-color . "white") (background-color . "DodgerBlue4") ;; (cursor-type . bar) (cursor-type . box) @end group @group (tool-bar-lines . 0) (menu-bar-lines . 1) (width . 80) (height . 58) (font . "-Misc-Fixed-Medium-R-Normal--20-200-75-75-C-100-ISO8859-1") )) @end group @end smallexample @item Convert @kbd{@key{CTRL}-h} into @key{DEL} and @key{DEL} into @kbd{@key{CTRL}-h}.@* (Some older keyboards needed this, although I have not seen the problem recently.) @smallexample @group ;; Translate 'C-h' to . ; (keyboard-translate ?\C-h ?\C-?) ;; Translate to 'C-h'. (keyboard-translate ?\C-? ?\C-h) @end group @end smallexample @item Turn off a blinking cursor! @smallexample @group (if (fboundp 'blink-cursor-mode) (blink-cursor-mode -1)) @end group @end smallexample @noindent or start GNU Emacs with the command @code{emacs -nbc}. @need 1250 @item When using @command{grep}@* @samp{-i}@w{ } Ignore case distinctions@* @samp{-n}@w{ } Prefix each line of output with line number@* @samp{-H}@w{ } Print the filename for each match.@* @samp{-e}@w{ } Protect patterns beginning with a hyphen character, @samp{-} @smallexample (setq grep-command "grep -i -nH -e ") @end smallexample @ignore @c Evidently, no longer needed in GNU Emacs 22 item Automatically uncompress compressed files when visiting them smallexample (load "uncompress") end smallexample @end ignore @item Find an existing buffer, even if it has a different name@* This avoids problems with symbolic links. @smallexample (setq find-file-existing-other-name t) @end smallexample @item Set your language environment and default input method @smallexample @group (set-language-environment "latin-1") ;; Remember you can enable or disable multilingual text input ;; with the @code{toggle-input-method'} (@kbd{C-\}) command (setq default-input-method "latin-1-prefix") @end group @end smallexample If you want to write with Chinese GB characters, set this instead: @smallexample @group (set-language-environment "Chinese-GB") (setq default-input-method "chinese-tonepy") @end group @end smallexample @end itemize @subsubheading Fixing Unpleasant Key Bindings @cindex Key bindings, fixing @cindex Bindings, key, fixing unpleasant Some systems bind keys unpleasantly. Sometimes, for example, the @key{CTRL} key appears in an awkward spot rather than at the far left of the home row. Usually, when people fix these sorts of keybindings, they do not change their @file{~/.emacs} file. Instead, they bind the proper keys on their consoles with the @code{loadkeys} or @code{install-keymap} commands in their boot script and then include @code{xmodmap} commands in their @file{.xinitrc} or @file{.Xsession} file for X Windows. @need 1250 @noindent For a boot script: @smallexample @group loadkeys /usr/share/keymaps/i386/qwerty/emacs2.kmap.gz @exdent or install-keymap emacs2 @end group @end smallexample @need 1250 @noindent For a @file{.xinitrc} or @file{.Xsession} file when the @key{Caps Lock} key is at the far left of the home row: @smallexample @group # Bind the key labeled 'Caps Lock' to 'Control' # (Such a broken user interface suggests that keyboard manufacturers # think that computers are typewriters from 1885.) xmodmap -e "clear Lock" xmodmap -e "add Control = Caps_Lock" @end group @end smallexample @need 1250 @noindent In a @file{.xinitrc} or @file{.Xsession} file, to convert an @key{ALT} key to a @key{META} key: @smallexample @group # Some ill designed keyboards have a key labeled ALT and no Meta xmodmap -e "keysym Alt_L = Meta_L Alt_L" @end group @end smallexample @need 1700 @node Mode Line @section A Modified Mode Line @vindex mode-line-format @cindex Mode line format Finally, a feature I really like: a modified mode line. When I work over a network, I forget which machine I am using. Also, I tend to I lose track of where I am, and which line point is on. So I reset my mode line to look like this: @smallexample -:-- foo.texi rattlesnake:/home/bob/ Line 1 (Texinfo Fill) Top @end smallexample I am visiting a file called @file{foo.texi}, on my machine @file{rattlesnake} in my @file{/home/bob} buffer. I am on line 1, in Texinfo mode, and am at the top of the buffer. @need 1200 My @file{.emacs} file has a section that looks like this: @smallexample @group ;; Set a Mode Line that tells me which machine, which directory, ;; and which line I am on, plus the other customary information. (setq-default mode-line-format (quote (#("-" 0 1 (help-echo "mouse-1: select window, mouse-2: delete others ...")) mode-line-mule-info mode-line-modified mode-line-frame-identification " " @end group @group mode-line-buffer-identification " " (:eval (substring (system-name) 0 (string-match "\\..+" (system-name)))) ":" default-directory #(" " 0 1 (help-echo "mouse-1: select window, mouse-2: delete others ...")) (line-number-mode " Line %l ") global-mode-string @end group @group #(" %[(" 0 6 (help-echo "mouse-1: select window, mouse-2: delete others ...")) (:eval (format-time-string "%F")) mode-line-process minor-mode-alist #("%n" 0 2 (help-echo "mouse-2: widen" local-map (keymap ...))) ")%] " (-3 . "%P") ;; "-%-" ))) @end group @end smallexample @noindent Here, I redefine the default mode line. Most of the parts are from the original; but I make a few changes. I set the @emph{default} mode line format so as to permit various modes, such as Info, to override it. Many elements in the list are self-explanatory: @code{mode-line-modified} is a variable that tells whether the buffer has been modified, @code{mode-name} tells the name of the mode, and so on. However, the format looks complicated because of two features we have not discussed. @cindex Properties, in mode line example The first string in the mode line is a dash or hyphen, @samp{-}. In the old days, it would have been specified simply as @code{"-"}. But nowadays, Emacs can add properties to a string, such as highlighting or, as in this case, a help feature. If you place your mouse cursor over the hyphen, some help information appears (By default, you must wait seven-tenths of a second before the information appears. You can change that timing by changing the value of @code{tooltip-delay}.) @need 1000 The new string format has a special syntax: @smallexample #("-" 0 1 (help-echo "mouse-1: select window, ...")) @end smallexample @noindent The @code{#(} begins a list. The first element of the list is the string itself, just one @samp{-}. The second and third elements specify the range over which the fourth element applies. A range starts @emph{after} a character, so a zero means the range starts just before the first character; a 1 means that the range ends just after the first character. The third element is the property for the range. It consists of a property list, a property name, in this case, @samp{help-echo}, followed by a value, in this case, a string. The second, third, and fourth elements of this new string format can be repeated. @xref{Text Properties, , Text Properties, elisp, The GNU Emacs Lisp Reference Manual}, and see @ref{Mode Line Format, , Mode Line Format, elisp, The GNU Emacs Lisp Reference Manual}, for more information. @code{mode-line-buffer-identification} displays the current buffer name. It is a list beginning @code{(#("%12b" 0 4 @dots{}}. The @code{#(} begins the list. The @samp{"%12b"} displays the current buffer name, using the @code{buffer-name} function with which we are familiar; the @samp{12} specifies the maximum number of characters that will be displayed. When a name has fewer characters, whitespace is added to fill out to this number. (Buffer names can and often should be longer than 12 characters; this length works well in a typical 80 column wide window.) @code{:eval} says to evaluate the following form and use the result as a string to display. In this case, the expression displays the first component of the full system name. The end of the first component is a @samp{.} (period), so I use the @code{string-match} function to tell me the length of the first component. The substring from the zeroth character to that length is the name of the machine. @need 1250 This is the expression: @smallexample @group (:eval (substring (system-name) 0 (string-match "\\..+" (system-name)))) @end group @end smallexample @samp{%[} and @samp{%]} cause a pair of square brackets to appear for each recursive editing level. @samp{%n} says ``Narrow'' when narrowing is in effect. @samp{%P} tells you the percentage of the buffer that is above the bottom of the window, or ``Top'', ``Bottom'', or ``All''. (A lower case @samp{p} tell you the percentage above the @emph{top} of the window.) @samp{%-} inserts enough dashes to fill out the line. Remember, you don't have to like Emacs to like it---your own Emacs can have different colors, different commands, and different keys than a default Emacs. On the other hand, if you want to bring up a plain out-of-the-box Emacs, with no customization, type: @smallexample emacs -q @end smallexample @noindent This will start an Emacs that does @emph{not} load your @file{~/.emacs} initialization file. A plain, default Emacs. Nothing more. @node Debugging @chapter Debugging @cindex debugging GNU Emacs has two debuggers, @code{debug} and @code{edebug}. The first is built into the internals of Emacs and is always with you; the second requires that you instrument a function before you can use it. Both debuggers are described extensively in @ref{Debugging, , Debugging Lisp Programs, elisp, The GNU Emacs Lisp Reference Manual}. In this chapter, I will walk through a short example of each. @menu * debug:: How to use the built-in debugger. * debug-on-entry:: Start debugging when you call a function. * debug-on-quit:: Start debugging when you quit with @kbd{C-g}. * edebug:: How to use Edebug, a source level debugger. * Debugging Exercises:: @end menu @node debug @section @code{debug} @findex debug Suppose you have written a function definition that is intended to return the sum of the numbers 1 through a given number. (This is the @code{triangle} function discussed earlier. @xref{Decrementing Example, , Example with Decrementing Counter}, for a discussion.) @c xref{Decrementing Loop,, Loop with a Decrementing Counter}, for a discussion.) However, your function definition has a bug. You have mistyped @samp{1=} for @samp{1-}. Here is the broken definition: @findex triangle-bugged @smallexample @group (defun triangle-bugged (number) "Return sum of numbers 1 through NUMBER inclusive." (let ((total 0)) (while (> number 0) (setq total (+ total number)) (setq number (1= number))) ; @r{Error here.} total)) @end group @end smallexample If you are reading this in Info, you can evaluate this definition in the normal fashion. You will see @code{triangle-bugged} appear in the echo area. @need 1250 Now evaluate the @code{triangle-bugged} function with an argument of 4: @smallexample (triangle-bugged 4) @end smallexample @noindent In a recent GNU Emacs, you will create and enter a @file{*Backtrace*} buffer that says: @noindent @smallexample @group ---------- Buffer: *Backtrace* ---------- Debugger entered--Lisp error: (void-function 1=) (1= number) (setq number (1= number)) (while (> number 0) (setq total (+ total number)) (setq number (1= number))) (let ((total 0)) (while (> number 0) (setq total ...) (setq number ...)) total) triangle-bugged(4) @end group @group eval((triangle-bugged 4) nil) eval-expression((triangle-bugged 4) nil nil 127) funcall-interactively(eval-expression (triangle-bugged 4) nil nil 127) call-interactively(eval-expression nil nil) command-execute(eval-expression) ---------- Buffer: *Backtrace* ---------- @end group @end smallexample @noindent (I have reformatted this example slightly; the debugger does not fold long lines. As usual, you can quit the debugger by typing @kbd{q} in the @file{*Backtrace*} buffer.) In practice, for a bug as simple as this, the Lisp error line will tell you what you need to know to correct the definition. The function @code{1=} is void. @ignore @need 800 In GNU Emacs 20 and before, you will see: @smallexample Symbol's function definition is void:@: 1= @end smallexample @noindent which has the same meaning as the @file{*Backtrace*} buffer line in version 21. @end ignore However, suppose you are not quite certain what is going on? You can read the complete backtrace. In this case, you need to run a recent GNU Emacs, which automatically starts the debugger that puts you in the @file{*Backtrace*} buffer; or else, you need to start the debugger manually as described below. Read the @file{*Backtrace*} buffer from the bottom up; it tells you what Emacs did that led to the error. Emacs made an interactive call to @kbd{C-x C-e} (@code{eval-last-sexp}), which led to the evaluation of the @code{triangle-bugged} expression. Each line above tells you what the Lisp interpreter evaluated next. @need 1250 The third line from the top of the buffer is @smallexample (setq number (1= number)) @end smallexample @noindent Emacs tried to evaluate this expression; in order to do so, it tried to evaluate the inner expression shown on the second line from the top: @smallexample (1= number) @end smallexample @need 1250 @noindent This is where the error occurred; as the top line says: @smallexample Debugger entered--Lisp error: (void-function 1=) @end smallexample @noindent You can correct the mistake, re-evaluate the function definition, and then run your test again. @node debug-on-entry @section @code{debug-on-entry} @findex debug-on-entry A recent GNU Emacs starts the debugger automatically when your function has an error. @ignore GNU Emacs version 20 and before did not; it simply presented you with an error message. You had to start the debugger manually. @end ignore Incidentally, you can start the debugger manually for all versions of Emacs; the advantage is that the debugger runs even if you do not have a bug in your code. Sometimes your code will be free of bugs! You can enter the debugger when you call the function by calling @code{debug-on-entry}. @need 1250 @noindent Type: @smallexample M-x debug-on-entry @key{RET} triangle-bugged @key{RET} @end smallexample @need 1250 @noindent Now, evaluate the following: @smallexample (triangle-bugged 5) @end smallexample @noindent All versions of Emacs will create a @file{*Backtrace*} buffer and tell you that it is beginning to evaluate the @code{triangle-bugged} function: @smallexample @group ---------- Buffer: *Backtrace* ---------- Debugger entered--entering a function: * triangle-bugged(5) eval((triangle-bugged 5) nil) @end group @group eval-expression((triangle-bugged 5) nil nil 127) funcall-interactively(eval-expression (triangle-bugged 5) nil nil 127) call-interactively(eval-expression nil nil) command-execute(eval-expression) ---------- Buffer: *Backtrace* ---------- @end group @end smallexample In the @file{*Backtrace*} buffer, type @kbd{d}. Emacs will evaluate the first expression in @code{triangle-bugged}; the buffer will look like this: @smallexample @group ---------- Buffer: *Backtrace* ---------- Debugger entered--beginning evaluation of function call form: * (let ((total 0)) (while (> number 0) (setq total ...) (setq number ...)) total) * triangle-bugged(5) eval((triangle-bugged 5)) @end group @group eval((triangle-bugged 5) nil) eval-expression((triangle-bugged 5) nil nil 127) funcall-interactively(eval-expression (triangle-bugged 5) nil nil 127) call-interactively(eval-expression nil nil) command-execute(eval-expression) ---------- Buffer: *Backtrace* ---------- @end group @end smallexample @noindent Now, type @kbd{d} again, eight times, slowly. Each time you type @kbd{d}, Emacs will evaluate another expression in the function definition. @need 1750 Eventually, the buffer will look like this: @smallexample @group ---------- Buffer: *Backtrace* ---------- Debugger entered--beginning evaluation of function call form: * (setq number (1= number)) * (while (> number 0) (setq total (+ total number)) (setq number (1= number))) @group @end group * (let ((total 0)) (while (> number 0) (setq total ...) (setq number ...)) total) * triangle-bugged(5) eval((triangle-bugged 5) nil) @group @end group eval-expression((triangle-bugged 5) nil nil 127) funcall-interactively(eval-expression (triangle-bugged 5) nil nil 127) call-interactively(eval-expression nil nil) command-execute(eval-expression) ---------- Buffer: *Backtrace* ---------- @end group @end smallexample @need 1500 @noindent Finally, after you type @kbd{d} two more times, Emacs will reach the error, and the top two lines of the @file{*Backtrace*} buffer will look like this: @smallexample @group ---------- Buffer: *Backtrace* ---------- Debugger entered--Lisp error: (void-function 1=) * (1= number) @dots{} ---------- Buffer: *Backtrace* ---------- @end group @end smallexample By typing @kbd{d}, you were able to step through the function. You can quit a @file{*Backtrace*} buffer by typing @kbd{q} in it; this quits the trace, but does not cancel @code{debug-on-entry}. @findex cancel-debug-on-entry To cancel the effect of @code{debug-on-entry}, call @code{cancel-debug-on-entry} and the name of the function, like this: @smallexample M-x cancel-debug-on-entry @key{RET} triangle-bugged @key{RET} @end smallexample @noindent (If you are reading this in Info, cancel @code{debug-on-entry} now.) @node debug-on-quit @section @code{debug-on-quit} and @code{(debug)} In addition to setting @code{debug-on-error} or calling @code{debug-on-entry}, there are two other ways to start @code{debug}. @findex debug-on-quit You can start @code{debug} whenever you type @kbd{C-g} (@code{keyboard-quit}) by setting the variable @code{debug-on-quit} to @code{t}. This is useful for debugging infinite loops. @need 1500 @cindex @code{(debug)} in code Or, you can insert a line that says @code{(debug)} into your code where you want the debugger to start, like this: @smallexample @group (defun triangle-bugged (number) "Return sum of numbers 1 through NUMBER inclusive." (let ((total 0)) (while (> number 0) (setq total (+ total number)) (debug) ; @r{Start debugger.} (setq number (1= number))) ; @r{Error here.} total)) @end group @end smallexample The @code{debug} function is described in detail in @ref{Debugger, , The Lisp Debugger, elisp, The GNU Emacs Lisp Reference Manual}. @node edebug @section The @code{edebug} Source Level Debugger @cindex Source level debugger @findex edebug Edebug is a source level debugger. Edebug normally displays the source of the code you are debugging, with an arrow at the left that shows which line you are currently executing. You can walk through the execution of a function, line by line, or run quickly until reaching a @dfn{breakpoint} where execution stops. Edebug is described in @ref{Edebug, , , elisp, The GNU Emacs Lisp Reference Manual}. @need 1250 Here is a bugged function definition for @code{triangle-recursively}. @xref{Recursive triangle function, , Recursion in place of a counter}, for a review of it. @smallexample @group (defun triangle-recursively-bugged (number) "Return sum of numbers 1 through NUMBER inclusive. Uses recursion." (if (= number 1) 1 (+ number (triangle-recursively-bugged (1= number))))) ; @r{Error here.} @end group @end smallexample @noindent Normally, you would install this definition by positioning your cursor after the function's closing parenthesis and typing @kbd{C-x C-e} (@code{eval-last-sexp}) or else by positioning your cursor within the definition and typing @kbd{C-M-x} (@code{eval-defun}). (By default, the @code{eval-defun} command works only in Emacs Lisp mode or in Lisp Interaction mode.) @need 1500 However, to prepare this function definition for Edebug, you must first @dfn{instrument} the code using a different command. You can do this by positioning your cursor within or just after the definition and typing @smallexample M-x edebug-defun @key{RET} @end smallexample @noindent This will cause Emacs to load Edebug automatically if it is not already loaded, and properly instrument the function. After instrumenting the function, place your cursor after the following expression and type @kbd{C-x C-e} (@code{eval-last-sexp}): @smallexample (triangle-recursively-bugged 3) @end smallexample @noindent You will be jumped back to the source for @code{triangle-recursively-bugged} and the cursor positioned at the beginning of the @code{if} line of the function. Also, you will see an arrowhead at the left hand side of that line. The arrowhead marks the line where the function is executing. (In the following examples, we show the arrowhead with @samp{=>}; in a windowing system, you may see the arrowhead as a solid triangle in the window fringe.) @smallexample =>@point{}(if (= number 1) @end smallexample @noindent @iftex In the example, the location of point is displayed with a star, @samp{@point{}} (in Info, it is displayed as @samp{-!-}). @end iftex @ifnottex In the example, the location of point is displayed as @samp{@point{}} (in a printed book, it is displayed with a five pointed star). @end ifnottex If you now press @key{SPC}, point will move to the next expression to be executed; the line will look like this: @smallexample =>(if @point{}(= number 1) @end smallexample @noindent As you continue to press @key{SPC}, point will move from expression to expression. At the same time, whenever an expression returns a value, that value will be displayed in the echo area. For example, after you move point past @code{number}, you will see the following: @smallexample Result: 3 (#o3, #x3, ?\C-c) @end smallexample @noindent This means the value of @code{number} is 3, which is octal three, hexadecimal three, and @sc{ascii} Control-C (the third letter of the alphabet, in case you need to know this information). You can continue moving through the code until you reach the line with the error. Before evaluation, that line looks like this: @smallexample => @point{}(1= number))))) ; @r{Error here.} @end smallexample @need 1250 @noindent When you press @key{SPC} once again, you will produce an error message that says: @smallexample Symbol's function definition is void:@: 1= @end smallexample @noindent This is the bug. Press @kbd{q} to quit Edebug. To remove instrumentation from a function definition, simply re-evaluate it with a command that does not instrument it. For example, you could place your cursor after the definition's closing parenthesis and type @kbd{C-x C-e}. Edebug does a great deal more than walk with you through a function. You can set it so it races through on its own, stopping only at an error or at specified stopping points; you can cause it to display the changing values of various expressions; you can find out how many times a function is called, and more. Edebug is described in @ref{Edebug, , , elisp, The GNU Emacs Lisp Reference Manual}. @need 1500 @node Debugging Exercises @section Debugging Exercises @itemize @bullet @item Install the @code{@value{COUNT-WORDS}} function and then cause it to enter the built-in debugger when you call it. Run the command on a region containing two words. You will need to press @kbd{d} a remarkable number of times. On your system, is a hook called after the command finishes? (For information on hooks, see @ref{Command Overview, , Command Loop Overview, elisp, The GNU Emacs Lisp Reference Manual}.) @item Copy @code{@value{COUNT-WORDS}} into the @file{*scratch*} buffer, instrument the function for Edebug, and walk through its execution. The function does not need to have a bug, although you can introduce one if you wish. If the function lacks a bug, the walk-through completes without problems. @item While running Edebug, type @kbd{?} to see a list of all the Edebug commands. (The @code{global-edebug-prefix} is usually @kbd{C-x X}, i.e., @kbd{@key{CTRL}-x} followed by an upper case @kbd{X}; use this prefix for commands made outside of the Edebug debugging buffer.) @item In the Edebug debugging buffer, use the @kbd{p} (@code{edebug-bounce-point}) command to see where in the region the @code{@value{COUNT-WORDS}} is working. @item Move point to some spot further down the function and then type the @kbd{h} (@code{edebug-goto-here}) command to jump to that location. @item Use the @kbd{t} (@code{edebug-trace-mode}) command to cause Edebug to walk through the function on its own; use an upper case @kbd{T} for @code{edebug-Trace-fast-mode}. @item Set a breakpoint, then run Edebug in Trace mode until it reaches the stopping point. @end itemize @node Conclusion @chapter Conclusion We have now reached the end of this Introduction. You have now learned enough about programming in Emacs Lisp to set values, to write simple @file{.emacs} files for yourself and your friends, and write simple customizations and extensions to Emacs. This is a place to stop. Or, if you wish, you can now go onward, and teach yourself. You have learned some of the basic nuts and bolts of programming. But only some. There are a great many more brackets and hinges that are easy to use that we have not touched. A path you can follow right now lies among the sources to GNU Emacs and in @ifnotinfo @cite{The GNU Emacs Lisp Reference Manual}. @end ifnotinfo @ifinfo @ref{Top, , The GNU Emacs Lisp Reference Manual, elisp, The GNU Emacs Lisp Reference Manual}. @end ifinfo The Emacs Lisp sources are an adventure. When you read the sources and come across a function or expression that is unfamiliar, you need to figure out or find out what it does. Go to the Reference Manual. It is a thorough, complete, and fairly easy-to-read description of Emacs Lisp. It is written not only for experts, but for people who know what you know. (The @cite{Reference Manual} comes with the standard GNU Emacs distribution. Like this introduction, it comes as a Texinfo source file, so you can read it on your computer and as a typeset, printed book.) Go to the other built-in help that is part of GNU Emacs: the built-in documentation for all functions and variables, and @code{xref-find-definitions}, the program that takes you to sources. Here is an example of how I explore the sources. Because of its name, @file{simple.el} is the file I looked at first, a long time ago. As it happens some of the functions in @file{simple.el} are complicated, or at least look complicated at first sight. The @code{open-line} function, for example, looks complicated. You may want to walk through this function slowly, as we did with the @code{forward-sentence} function. (@xref{forward-sentence, The @code{forward-sentence} function}.) Or you may want to skip that function and look at another, such as @code{split-line}. You don't need to read all the functions. According to @code{count-words-in-defun}, the @code{split-line} function contains 102 words and symbols. Even though it is short, @code{split-line} contains expressions we have not studied: @code{skip-chars-forward}, @code{indent-to}, @code{current-column} and @code{insert-and-inherit}. Consider the @code{skip-chars-forward} function. In GNU Emacs, you can find out more about @code{skip-chars-forward} by typing @kbd{C-h f} (@code{describe-function}) and the name of the function. This gives you the function documentation. You may be able to guess what is done by a well named function such as @code{indent-to}; or you can look it up, too. Incidentally, the @code{describe-function} function itself is in @file{help.el}; it is one of those long, but decipherable functions. You can look up @code{describe-function} using the @kbd{C-h f} command! In this instance, since the code is Lisp, the @file{*Help*} buffer contains the name of the library containing the function's source. You can put point over the name of the library and press the @key{RET} key, which in this situation is bound to @code{help-follow}, and be taken directly to the source, in the same way as @kbd{M-.} (@code{xref-find-definitions}). The definition for @code{describe-function} illustrates how to customize the @code{interactive} expression without using the standard character codes; and it shows how to create a temporary buffer. (The @code{indent-to} function is written in C rather than Emacs Lisp; it is a built-in function. @code{help-follow} takes you to its source as does @code{xref-find-definitions}, when properly set up.) You can look at a function's source using @code{xref-find-definitions}, which is bound to @kbd{M-.} Finally, you can find out what the Reference Manual has to say by visiting the manual in Info, and typing @kbd{i} (@code{Info-index}) and the name of the function, or by looking up the function in the index to a printed copy of the manual. Similarly, you can find out what is meant by @code{insert-and-inherit}. Other interesting source files include @file{paragraphs.el}, @file{loaddefs.el}, and @file{loadup.el}. The @file{paragraphs.el} file includes short, easily understood functions as well as longer ones. The @file{loaddefs.el} file contains the many standard autoloads and many keymaps. I have never looked at it all; only at parts. @file{loadup.el} is the file that loads the standard parts of Emacs; it tells you a great deal about how Emacs is built. (@xref{Building Emacs, , Building Emacs, elisp, The GNU Emacs Lisp Reference Manual}, for more about building.) As I said, you have learned some nuts and bolts; however, and very importantly, we have hardly touched major aspects of programming; I have said nothing about how to sort information, except to use the predefined @code{sort} function; I have said nothing about how to store information, except to use variables and lists; I have said nothing about how to write programs that write programs. These are topics for another, and different kind of book, a different kind of learning. What you have done is learn enough for much practical work with GNU Emacs. What you have done is get started. This is the end of a beginning. @c ================ Appendix ================ @node the-the @appendix The @code{the-the} Function @findex the-the @cindex Duplicated words function @cindex Words, duplicated Sometimes when you you write text, you duplicate words---as with ``you you'' near the beginning of this sentence. I find that most frequently, I duplicate ``the''; hence, I call the function for detecting duplicated words, @code{the-the}. @need 1250 As a first step, you could use the following regular expression to search for duplicates: @smallexample \\(\\w+[ \t\n]+\\)\\1 @end smallexample @noindent This regexp matches one or more word-constituent characters followed by one or more spaces, tabs, or newlines. However, it does not detect duplicated words on different lines, since the ending of the first word, the end of the line, is different from the ending of the second word, a space. (For more information about regular expressions, see @ref{Regexp Search, , Regular Expression Searches}, as well as @ref{Regexps, , Syntax of Regular Expressions, emacs, The GNU Emacs Manual}, and @ref{Regular Expressions, , Regular Expressions, elisp, The GNU Emacs Lisp Reference Manual}.) You might try searching just for duplicated word-constituent characters but that does not work since the pattern detects doubles such as the two occurrences of ``th'' in ``with the''. Another possible regexp searches for word-constituent characters followed by non-word-constituent characters, reduplicated. Here, @w{@samp{\\w+}} matches one or more word-constituent characters and @w{@samp{\\W*}} matches zero or more non-word-constituent characters. @smallexample \\(\\(\\w+\\)\\W*\\)\\1 @end smallexample @noindent Again, not useful. Here is the pattern that I use. It is not perfect, but good enough. @w{@samp{\\b}} matches the empty string, provided it is at the beginning or end of a word; @w{@samp{[^@@ \n\t]+}} matches one or more occurrences of any characters that are @emph{not} an @@-sign, space, newline, or tab. @smallexample \\b\\([^@@ \n\t]+\\)[ \n\t]+\\1\\b @end smallexample One can write more complicated expressions, but I found that this expression is good enough, so I use it. Here is the @code{the-the} function, as I include it in my @file{.emacs} file, along with a handy global key binding: @smallexample @group (defun the-the () "Search forward for for a duplicated word." (interactive) (message "Searching for for duplicated words ...") (push-mark) @end group @group ;; This regexp is not perfect ;; but is fairly good over all: (if (re-search-forward "\\b\\([^@@ \n\t]+\\)[ \n\t]+\\1\\b" nil 'move) (message "Found duplicated word.") (message "End of buffer"))) @end group @group ;; Bind 'the-the' to C-c \ (global-set-key "\C-c\\" 'the-the) @end group @end smallexample @sp 1 Here is test text: @smallexample @group one two two three four five five six seven @end group @end smallexample You can substitute the other regular expressions shown above in the function definition and try each of them on this list. @node Kill Ring @appendix Handling the Kill Ring @cindex Kill ring handling @cindex Handling the kill ring @cindex Ring, making a list like a The kill ring is a list that is transformed into a ring by the workings of the @code{current-kill} function. The @code{yank} and @code{yank-pop} commands use the @code{current-kill} function. This appendix describes the @code{current-kill} function as well as both the @code{yank} and the @code{yank-pop} commands, but first, consider the workings of the kill ring. @menu * What the Kill Ring Does:: * current-kill:: * yank:: Paste a copy of a clipped element. * yank-pop:: Insert element pointed to. * ring file:: @end menu @ifnottex @node What the Kill Ring Does @unnumberedsec What the Kill Ring Does @end ifnottex @need 1250 The kill ring has a default maximum length of sixty items; this number is too large for an explanation. Instead, set it to four. Please evaluate the following: @smallexample @group (setq old-kill-ring-max kill-ring-max) (setq kill-ring-max 4) @end group @end smallexample @noindent Then, please copy each line of the following indented example into the kill ring. You may kill each line with @kbd{C-k} or mark it and copy it with @kbd{M-w}. @noindent (In a read-only buffer, such as the @file{*info*} buffer, the kill command, @kbd{C-k} (@code{kill-line}), will not remove the text, merely copy it to the kill ring. However, your machine may beep at you. Alternatively, for silence, you may copy the region of each line with the @kbd{M-w} (@code{kill-ring-save}) command. You must mark each line for this command to succeed, but it does not matter at which end you put point or mark.) @need 1250 @noindent Please invoke the calls in order, so that five elements attempt to fill the kill ring: @smallexample @group first some text second piece of text third line fourth line of text fifth bit of text @end group @end smallexample @need 1250 @noindent Then find the value of @code{kill-ring} by evaluating @smallexample kill-ring @end smallexample @need 800 @noindent It is: @smallexample @group ("fifth bit of text" "fourth line of text" "third line" "second piece of text") @end group @end smallexample @noindent The first element, @samp{first some text}, was dropped. @need 1250 To return to the old value for the length of the kill ring, evaluate: @smallexample (setq kill-ring-max old-kill-ring-max) @end smallexample @node current-kill @appendixsec The @code{current-kill} Function @findex current-kill The @code{current-kill} function changes the element in the kill ring to which @code{kill-ring-yank-pointer} points. (Also, the @code{kill-new} function sets @code{kill-ring-yank-pointer} to point to the latest element of the kill ring. The @code{kill-new} function is used directly or indirectly by @code{kill-append}, @code{copy-region-as-kill}, @code{kill-ring-save}, @code{kill-line}, and @code{kill-region}.) @menu * Code for current-kill:: * Understanding current-kill:: @end menu @ifnottex @node Code for current-kill @unnumberedsubsec The code for @code{current-kill} @end ifnottex @need 1500 The @code{current-kill} function is used by @code{yank} and by @code{yank-pop}. Here is the code for @code{current-kill}: @smallexample @group (defun current-kill (n &optional do-not-move) "Rotate the yanking point by N places, and then return that kill. If N is zero and `interprogram-paste-function' is set to a function that returns a string or a list of strings, and if that function doesn't return nil, then that string (or list) is added to the front of the kill ring and the string (or first string in the list) is returned as the latest kill. @end group @group If N is not zero, and if `yank-pop-change-selection' is non-nil, use `interprogram-cut-function' to transfer the kill at the new yank point into the window system selection. @end group @group If optional arg DO-NOT-MOVE is non-nil, then don't actually move the yanking point; just return the Nth kill forward." (let ((interprogram-paste (and (= n 0) interprogram-paste-function (funcall interprogram-paste-function)))) @end group @group (if interprogram-paste (progn ;; Disable the interprogram cut function when we add the new ;; text to the kill ring, so Emacs doesn't try to own the ;; selection, with identical text. (let ((interprogram-cut-function nil)) (if (listp interprogram-paste) (mapc 'kill-new (nreverse interprogram-paste)) (kill-new interprogram-paste))) (car kill-ring)) @end group @group (or kill-ring (error "Kill ring is empty")) (let ((ARGth-kill-element (nthcdr (mod (- n (length kill-ring-yank-pointer)) (length kill-ring)) kill-ring))) (unless do-not-move (setq kill-ring-yank-pointer ARGth-kill-element) (when (and yank-pop-change-selection (> n 0) interprogram-cut-function) (funcall interprogram-cut-function (car ARGth-kill-element)))) (car ARGth-kill-element))))) @end group @end smallexample Remember also that the @code{kill-new} function sets @code{kill-ring-yank-pointer} to the latest element of the kill ring, which means that all the functions that call it set the value indirectly: @code{kill-append}, @code{copy-region-as-kill}, @code{kill-ring-save}, @code{kill-line}, and @code{kill-region}. @need 1500 Here is the line in @code{kill-new}, which is explained in @ref{kill-new function, , The @code{kill-new} function}. @smallexample (setq kill-ring-yank-pointer kill-ring) @end smallexample @ifnottex @node Understanding current-kill @unnumberedsubsec @code{current-kill} in Outline @end ifnottex The @code{current-kill} function looks complex, but as usual, it can be understood by taking it apart piece by piece. First look at it in skeletal form: @smallexample @group (defun current-kill (n &optional do-not-move) "Rotate the yanking point by N places, and then return that kill." (let @var{varlist} @var{body}@dots{}) @end group @end smallexample This function takes two arguments, one of which is optional. It has a documentation string. It is @emph{not} interactive. @menu * Body of current-kill:: * Digression concerning error:: How to mislead humans, but not computers. * Determining the Element:: @end menu @ifnottex @node Body of current-kill @unnumberedsubsubsec The Body of @code{current-kill} @end ifnottex The body of the function definition is a @code{let} expression, which itself has a body as well as a @var{varlist}. The @code{let} expression declares a variable that will be only usable within the bounds of this function. This variable is called @code{interprogram-paste} and is for copying to another program. It is not for copying within this instance of GNU Emacs. Most window systems provide a facility for interprogram pasting. Sadly, that facility usually provides only for the last element. Most windowing systems have not adopted a ring of many possibilities, even though Emacs has provided it for decades. The @code{if} expression has two parts, one if there exists @code{interprogram-paste} and one if not. @need 2000 Let us consider the else-part of the @code{current-kill} function. (The then-part uses the @code{kill-new} function, which we have already described. @xref{kill-new function, , The @code{kill-new} function}.) @smallexample @group (or kill-ring (error "Kill ring is empty")) (let ((ARGth-kill-element (nthcdr (mod (- n (length kill-ring-yank-pointer)) (length kill-ring)) kill-ring))) (or do-not-move (setq kill-ring-yank-pointer ARGth-kill-element)) (car ARGth-kill-element)) @end group @end smallexample @noindent The code first checks whether the kill ring has content; otherwise it signals an error. @need 1000 Note that the @code{or} expression is very similar to testing length with an @code{if}: @findex zerop @findex error @smallexample @group (if (zerop (length kill-ring)) ; @r{if-part} (error "Kill ring is empty")) ; @r{then-part} ;; No else-part @end group @end smallexample @noindent If there is not anything in the kill ring, its length must be zero and an error message sent to the user: @samp{Kill ring is empty}. The @code{current-kill} function uses an @code{or} expression which is simpler. But an @code{if} expression reminds us what goes on. This @code{if} expression uses the function @code{zerop} which returns true if the value it is testing is zero. When @code{zerop} tests true, the then-part of the @code{if} is evaluated. The then-part is a list starting with the function @code{error}, which is a function that is similar to the @code{message} function (@pxref{message, , The @code{message} Function}) in that it prints a one-line message in the echo area. However, in addition to printing a message, @code{error} also stops evaluation of the function within which it is embedded. This means that the rest of the function will not be evaluated if the length of the kill ring is zero. Then the @code{current-kill} function selects the element to return. The selection depends on the number of places that @code{current-kill} rotates and on where @code{kill-ring-yank-pointer} points. Next, either the optional @code{do-not-move} argument is true or the current value of @code{kill-ring-yank-pointer} is set to point to the list. Finally, another expression returns the first element of the list even if the @code{do-not-move} argument is true. @ifnottex @node Digression concerning error @unnumberedsubsubsec Digression about the word ``error'' @end ifnottex In my opinion, it is slightly misleading, at least to humans, to use the term ``error'' as the name of the @code{error} function. A better term would be ``cancel''. Strictly speaking, of course, you cannot point to, much less rotate a pointer to a list that has no length, so from the point of view of the computer, the word ``error'' is correct. But a human expects to attempt this sort of thing, if only to find out whether the kill ring is full or empty. This is an act of exploration. From the human point of view, the act of exploration and discovery is not necessarily an error, and therefore should not be labeled as one, even in the bowels of a computer. As it is, the code in Emacs implies that a human who is acting virtuously, by exploring his or her environment, is making an error. This is bad. Even though the computer takes the same steps as it does when there is an error, a term such as ``cancel'' would have a clearer connotation. @ifnottex @node Determining the Element @unnumberedsubsubsec Determining the Element @end ifnottex Among other actions, the else-part of the @code{if} expression sets the value of @code{kill-ring-yank-pointer} to @code{ARGth-kill-element} when the kill ring has something in it and the value of @code{do-not-move} is @code{nil}. @need 800 The code looks like this: @smallexample @group (nthcdr (mod (- n (length kill-ring-yank-pointer)) (length kill-ring)) kill-ring))) @end group @end smallexample This needs some examination. Unless it is not supposed to move the pointer, the @code{current-kill} function changes where @code{kill-ring-yank-pointer} points. That is what the @w{@code{(setq kill-ring-yank-pointer ARGth-kill-element))}} expression does. Also, clearly, @code{ARGth-kill-element} is being set to be equal to some @sc{cdr} of the kill ring, using the @code{nthcdr} function that is described in an earlier section. (@xref{copy-region-as-kill}.) How does it do this? As we have seen before (@pxref{nthcdr}), the @code{nthcdr} function works by repeatedly taking the @sc{cdr} of a list---it takes the @sc{cdr} of the @sc{cdr} of the @sc{cdr} @dots{} @need 800 The two following expressions produce the same result: @smallexample @group (setq kill-ring-yank-pointer (cdr kill-ring)) (setq kill-ring-yank-pointer (nthcdr 1 kill-ring)) @end group @end smallexample However, the @code{nthcdr} expression is more complicated. It uses the @code{mod} function to determine which @sc{cdr} to select. (You will remember to look at inner functions first; indeed, we will have to go inside the @code{mod}.) The @code{mod} function returns the value of its first argument modulo the second; that is to say, it returns the remainder after dividing the first argument by the second. The value returned has the same sign as the second argument. @need 800 Thus, @smallexample @group (mod 12 4) @result{} 0 ;; @r{because there is no remainder} (mod 13 4) @result{} 1 @end group @end smallexample @need 1250 In this case, the first argument is often smaller than the second. That is fine. @smallexample @group (mod 0 4) @result{} 0 (mod 1 4) @result{} 1 @end group @end smallexample We can guess what the @code{-} function does. It is like @code{+} but subtracts instead of adds; the @code{-} function subtracts its second argument from its first. Also, we already know what the @code{length} function does (@pxref{length}). It returns the length of a list. And @code{n} is the name of the required argument to the @code{current-kill} function. @need 1250 So when the first argument to @code{nthcdr} is zero, the @code{nthcdr} expression returns the whole list, as you can see by evaluating the following: @smallexample @group ;; kill-ring-yank-pointer @r{and} kill-ring @r{have a length of four} ;; @r{and} (mod (- 0 4) 4) @result{} 0 (nthcdr (mod (- 0 4) 4) '("fourth line of text" "third line" "second piece of text" "first some text")) @end group @end smallexample @need 1250 When the first argument to the @code{current-kill} function is one, the @code{nthcdr} expression returns the list without its first element. @smallexample @group (nthcdr (mod (- 1 4) 4) '("fourth line of text" "third line" "second piece of text" "first some text")) @end group @end smallexample @cindex @samp{global variable} defined @cindex @samp{variable, global}, defined Incidentally, both @code{kill-ring} and @code{kill-ring-yank-pointer} are @dfn{global variables}. That means that any expression in Emacs Lisp can access them. They are not like the local variables set by @code{let} or like the symbols in an argument list. Local variables can only be accessed within the @code{let} that defines them or the function that specifies them in an argument list (and within expressions called by them). @c texi2dvi fails when the name of the section is within ifnottex ... @ifnottex (@xref{Prevent confusion, , @code{let} Prevents Confusion}, and @end ifnottex @iftex (@xref{Permanent Installation, , @code{let} Prevents Confusion}, and @end iftex @ref{defun, , The @code{defun} Macro}.) @node yank @appendixsec @code{yank} @findex yank After learning about @code{current-kill}, the code for the @code{yank} function is almost easy. The @code{yank} function does not use the @code{kill-ring-yank-pointer} variable directly. It calls @code{insert-for-yank} which calls @code{current-kill} which sets the @code{kill-ring-yank-pointer} variable. @need 1250 The code looks like this: @c in GNU Emacs 22 @smallexample @group (defun yank (&optional arg) "Reinsert (\"paste\") the last stretch of killed text. More precisely, reinsert the stretch of killed text most recently killed OR yanked. Put point at end, and set mark at beginning. With just \\[universal-argument] as argument, same but put point at beginning (and mark at end). With argument N, reinsert the Nth most recently killed stretch of killed text. When this command inserts killed text into the buffer, it honors `yank-excluded-properties' and `yank-handler' as described in the doc string for `insert-for-yank-1', which see. See also the command `yank-pop' (\\[yank-pop])." @end group @group (interactive "*P") (setq yank-window-start (window-start)) ;; If we don't get all the way thru, make last-command indicate that ;; for the following command. (setq this-command t) (push-mark (point)) @end group @group (insert-for-yank (current-kill (cond ((listp arg) 0) ((eq arg '-) -2) (t (1- arg))))) (if (consp arg) ;; This is like exchange-point-and-mark, but doesn't activate the mark. ;; It is cleaner to avoid activation, even though the command ;; loop would deactivate the mark because we inserted text. (goto-char (prog1 (mark t) (set-marker (mark-marker) (point) (current-buffer))))) @end group @group ;; If we do get all the way thru, make this-command indicate that. (if (eq this-command t) (setq this-command 'yank)) nil) @end group @end smallexample The key expression is @code{insert-for-yank}, which inserts the string returned by @code{current-kill}, but removes some text properties from it. However, before getting to that expression, the function sets the value of @code{yank-window-start} to the position returned by the @code{(window-start)} expression, the position at which the display currently starts. The @code{yank} function also sets @code{this-command} and pushes the mark. After it yanks the appropriate element, if the optional argument is a @sc{cons} rather than a number or nothing, it puts point at beginning of the yanked text and mark at its end. (The @code{prog1} function is like @code{progn} but returns the value of its first argument rather than the value of its last argument. Its first argument is forced to return the buffer's mark as an integer. You can see the documentation for these functions by placing point over them in this buffer and then typing @kbd{C-h f} (@code{describe-function}) followed by a @kbd{RET}; the default is the function.) The last part of the function tells what to do when it succeeds. @node yank-pop @appendixsec @code{yank-pop} @findex yank-pop After understanding @code{yank} and @code{current-kill}, you know how to approach the @code{yank-pop} function. Leaving out the documentation to save space, it looks like this: @c GNU Emacs 22 @smallexample @group (defun yank-pop (&optional arg) "@dots{}" (interactive "*p") (if (not (eq last-command 'yank)) (error "Previous command was not a yank")) @end group @group (setq this-command 'yank) (unless arg (setq arg 1)) (let ((inhibit-read-only t) (before (< (point) (mark t)))) @end group @group (if before (funcall (or yank-undo-function 'delete-region) (point) (mark t)) (funcall (or yank-undo-function 'delete-region) (mark t) (point))) (setq yank-undo-function nil) @end group @group (set-marker (mark-marker) (point) (current-buffer)) (insert-for-yank (current-kill arg)) ;; Set the window start back where it was in the yank command, ;; if possible. (set-window-start (selected-window) yank-window-start t) @end group @group (if before ;; This is like exchange-point-and-mark, ;; but doesn't activate the mark. ;; It is cleaner to avoid activation, even though the command ;; loop would deactivate the mark because we inserted text. (goto-char (prog1 (mark t) (set-marker (mark-marker) (point) (current-buffer)))))) nil) @end group @end smallexample The function is interactive with a small @samp{p} so the prefix argument is processed and passed to the function. The command can only be used after a previous yank; otherwise an error message is sent. This check uses the variable @code{last-command} which is set by @code{yank} and is discussed elsewhere. (@xref{copy-region-as-kill}.) The @code{let} clause sets the variable @code{before} to true or false depending whether point is before or after mark and then the region between point and mark is deleted. This is the region that was just inserted by the previous yank and it is this text that will be replaced. @code{funcall} calls its first argument as a function, passing remaining arguments to it. The first argument is whatever the @code{or} expression returns. The two remaining arguments are the positions of point and mark set by the preceding @code{yank} command. There is more, but that is the hardest part. @node ring file @appendixsec The @file{ring.el} File @cindex @file{ring.el} file Interestingly, GNU Emacs posses a file called @file{ring.el} that provides many of the features we just discussed. But functions such as @code{kill-ring-yank-pointer} do not use this library, possibly because they were written earlier. @node Full Graph @appendix A Graph with Labeled Axes Printed axes help you understand a graph. They convey scale. In an earlier chapter (@pxref{Readying a Graph, , Readying a Graph}), we wrote the code to print the body of a graph. Here we write the code for printing and labeling vertical and horizontal axes, along with the body itself. @menu * Labeled Example:: * print-graph Varlist:: @code{let} expression in @code{print-graph}. * print-Y-axis:: Print a label for the vertical axis. * print-X-axis:: Print a horizontal label. * Print Whole Graph:: The function to print a complete graph. @end menu @ifnottex @node Labeled Example @unnumberedsec Labeled Example Graph @end ifnottex Since insertions fill a buffer to the right and below point, the new graph printing function should first print the Y or vertical axis, then the body of the graph, and finally the X or horizontal axis. This sequence lays out for us the contents of the function: @enumerate @item Set up code. @item Print Y axis. @item Print body of graph. @item Print X axis. @end enumerate @need 800 Here is an example of how a finished graph should look: @smallexample @group 10 - * * * * ** * *** 5 - * ******* * *** ******* ************* *************** 1 - **************** | | | | 1 5 10 15 @end group @end smallexample @noindent In this graph, both the vertical and the horizontal axes are labeled with numbers. However, in some graphs, the horizontal axis is time and would be better labeled with months, like this: @smallexample @group 5 - * * ** * ******* ********** ** 1 - ************** | ^ | Jan June Jan @end group @end smallexample Indeed, with a little thought, we can easily come up with a variety of vertical and horizontal labeling schemes. Our task could become complicated. But complications breed confusion. Rather than permit this, it is better choose a simple labeling scheme for our first effort, and to modify or replace it later. @need 1200 These considerations suggest the following outline for the @code{print-graph} function: @smallexample @group (defun print-graph (numbers-list) "@var{documentation}@dots{}" (let ((height @dots{} @dots{})) @end group @group (print-Y-axis height @dots{} ) (graph-body-print numbers-list) (print-X-axis @dots{} ))) @end group @end smallexample We can work on each part of the @code{print-graph} function definition in turn. @node print-graph Varlist @appendixsec The @code{print-graph} Varlist @cindex @code{print-graph} varlist In writing the @code{print-graph} function, the first task is to write the varlist in the @code{let} expression. (We will leave aside for the moment any thoughts about making the function interactive or about the contents of its documentation string.) The varlist should set several values. Clearly, the top of the label for the vertical axis must be at least the height of the graph, which means that we must obtain this information here. Note that the @code{print-graph-body} function also requires this information. There is no reason to calculate the height of the graph in two different places, so we should change @code{print-graph-body} from the way we defined it earlier to take advantage of the calculation. Similarly, both the function for printing the X axis labels and the @code{print-graph-body} function need to learn the value of the width of each symbol. We can perform the calculation here and change the definition for @code{print-graph-body} from the way we defined it in the previous chapter. The length of the label for the horizontal axis must be at least as long as the graph. However, this information is used only in the function that prints the horizontal axis, so it does not need to be calculated here. These thoughts lead us directly to the following form for the varlist in the @code{let} for @code{print-graph}: @smallexample @group (let ((height (apply 'max numbers-list)) ; @r{First version.} (symbol-width (length graph-blank))) @end group @end smallexample @noindent As we shall see, this expression is not quite right. @need 2000 @node print-Y-axis @appendixsec The @code{print-Y-axis} Function @cindex Axis, print vertical @cindex Y axis printing @cindex Vertical axis printing @cindex Print vertical axis The job of the @code{print-Y-axis} function is to print a label for the vertical axis that looks like this: @smallexample @group 10 - 5 - 1 - @end group @end smallexample @noindent The function should be passed the height of the graph, and then should construct and insert the appropriate numbers and marks. @menu * print-Y-axis in Detail:: * Height of label:: What height for the Y axis? * Compute a Remainder:: How to compute the remainder of a division. * Y Axis Element:: Construct a line for the Y axis. * Y-axis-column:: Generate a list of Y axis labels. * print-Y-axis Penultimate:: A not quite final version. @end menu @ifnottex @node print-Y-axis in Detail @unnumberedsubsec The @code{print-Y-axis} Function in Detail @end ifnottex It is easy enough to see in the figure what the Y axis label should look like; but to say in words, and then to write a function definition to do the job is another matter. It is not quite true to say that we want a number and a tic every five lines: there are only three lines between the @samp{1} and the @samp{5} (lines 2, 3, and 4), but four lines between the @samp{5} and the @samp{10} (lines 6, 7, 8, and 9). It is better to say that we want a number and a tic mark on the base line (number 1) and then that we want a number and a tic on the fifth line from the bottom and on every line that is a multiple of five. @ifnottex @node Height of label @unnumberedsubsec What height should the label be? @end ifnottex The next issue is what height the label should be? Suppose the maximum height of tallest column of the graph is seven. Should the highest label on the Y axis be @samp{5 -}, and should the graph stick up above the label? Or should the highest label be @samp{7 -}, and mark the peak of the graph? Or should the highest label be @code{10 -}, which is a multiple of five, and be higher than the topmost value of the graph? The latter form is preferred. Most graphs are drawn within rectangles whose sides are an integral number of steps long---5, 10, 15, and so on for a step distance of five. But as soon as we decide to use a step height for the vertical axis, we discover that the simple expression in the varlist for computing the height is wrong. The expression is @code{(apply 'max numbers-list)}. This returns the precise height, not the maximum height plus whatever is necessary to round up to the nearest multiple of five. A more complex expression is required. As usual in cases like this, a complex problem becomes simpler if it is divided into several smaller problems. First, consider the case when the highest value of the graph is an integral multiple of five---when it is 5, 10, 15, or some higher multiple of five. We can use this value as the Y axis height. A fairly simply way to determine whether a number is a multiple of five is to divide it by five and see if the division results in a remainder. If there is no remainder, the number is a multiple of five. Thus, seven divided by five has a remainder of two, and seven is not an integral multiple of five. Put in slightly different language, more reminiscent of the classroom, five goes into seven once, with a remainder of two. However, five goes into ten twice, with no remainder: ten is an integral multiple of five. @node Compute a Remainder @appendixsubsec Side Trip: Compute a Remainder @findex % @r{(remainder function)} @cindex Remainder function, @code{%} In Lisp, the function for computing a remainder is @code{%}. The function returns the remainder of its first argument divided by its second argument. As it happens, @code{%} is a function in Emacs Lisp that you cannot discover using @code{apropos}: you find nothing if you type @kbd{M-x apropos @key{RET} remainder @key{RET}}. The only way to learn of the existence of @code{%} is to read about it in a book such as this or in the Emacs Lisp sources. You can try the @code{%} function by evaluating the following two expressions: @smallexample @group (% 7 5) (% 10 5) @end group @end smallexample @noindent The first expression returns 2 and the second expression returns 0. To test whether the returned value is zero or some other number, we can use the @code{zerop} function. This function returns @code{t} if its argument, which must be a number, is zero. @smallexample @group (zerop (% 7 5)) @result{} nil (zerop (% 10 5)) @result{} t @end group @end smallexample Thus, the following expression will return @code{t} if the height of the graph is evenly divisible by five: @smallexample (zerop (% height 5)) @end smallexample @noindent (The value of @code{height}, of course, can be found from @code{(apply 'max numbers-list)}.) On the other hand, if the value of @code{height} is not a multiple of five, we want to reset the value to the next higher multiple of five. This is straightforward arithmetic using functions with which we are already familiar. First, we divide the value of @code{height} by five to determine how many times five goes into the number. Thus, five goes into twelve twice. If we add one to this quotient and multiply by five, we will obtain the value of the next multiple of five that is larger than the height. Five goes into twelve twice. Add one to two, and multiply by five; the result is fifteen, which is the next multiple of five that is higher than twelve. The Lisp expression for this is: @smallexample (* (1+ (/ height 5)) 5) @end smallexample @noindent For example, if you evaluate the following, the result is 15: @smallexample (* (1+ (/ 12 5)) 5) @end smallexample All through this discussion, we have been using 5 as the value for spacing labels on the Y axis; but we may want to use some other value. For generality, we should replace 5 with a variable to which we can assign a value. The best name I can think of for this variable is @code{Y-axis-label-spacing}. @need 1250 Using this term, and an @code{if} expression, we produce the following: @smallexample @group (if (zerop (% height Y-axis-label-spacing)) height ;; @r{else} (* (1+ (/ height Y-axis-label-spacing)) Y-axis-label-spacing)) @end group @end smallexample @noindent This expression returns the value of @code{height} itself if the height is an even multiple of the value of the @code{Y-axis-label-spacing} or else it computes and returns a value of @code{height} that is equal to the next higher multiple of the value of the @code{Y-axis-label-spacing}. We can now include this expression in the @code{let} expression of the @code{print-graph} function (after first setting the value of @code{Y-axis-label-spacing}): @vindex Y-axis-label-spacing @smallexample @group (defvar Y-axis-label-spacing 5 "Number of lines from one Y axis label to next.") @end group @group @dots{} (let* ((height (apply 'max numbers-list)) (height-of-top-line (if (zerop (% height Y-axis-label-spacing)) height @end group @group ;; @r{else} (* (1+ (/ height Y-axis-label-spacing)) Y-axis-label-spacing))) (symbol-width (length graph-blank)))) @dots{} @end group @end smallexample @noindent (Note use of the @code{let*} function: the initial value of height is computed once by the @code{(apply 'max numbers-list)} expression and then the resulting value of @code{height} is used to compute its final value. @xref{fwd-para let, , The @code{let*} expression}, for more about @code{let*}.) @node Y Axis Element @appendixsubsec Construct a Y Axis Element When we print the vertical axis, we want to insert strings such as @w{@samp{5 -}} and @w{@samp{10 - }} every five lines. Moreover, we want the numbers and dashes to line up, so shorter numbers must be padded with leading spaces. If some of the strings use two digit numbers, the strings with single digit numbers must include a leading blank space before the number. @findex number-to-string To figure out the length of the number, the @code{length} function is used. But the @code{length} function works only with a string, not with a number. So the number has to be converted from being a number to being a string. This is done with the @code{number-to-string} function. For example, @smallexample @group (length (number-to-string 35)) @result{} 2 (length (number-to-string 100)) @result{} 3 @end group @end smallexample @noindent (@code{number-to-string} is also called @code{int-to-string}; you will see this alternative name in various sources.) In addition, in each label, each number is followed by a string such as @w{@samp{ - }}, which we will call the @code{Y-axis-tic} marker. This variable is defined with @code{defvar}: @vindex Y-axis-tic @smallexample @group (defvar Y-axis-tic " - " "String that follows number in a Y axis label.") @end group @end smallexample The length of the Y label is the sum of the length of the Y axis tic mark and the length of the number of the top of the graph. @smallexample (length (concat (number-to-string height) Y-axis-tic))) @end smallexample This value will be calculated by the @code{print-graph} function in its varlist as @code{full-Y-label-width} and passed on. (Note that we did not think to include this in the varlist when we first proposed it.) To make a complete vertical axis label, a tic mark is concatenated with a number; and the two together may be preceded by one or more spaces depending on how long the number is. The label consists of three parts: the (optional) leading spaces, the number, and the tic mark. The function is passed the value of the number for the specific row, and the value of the width of the top line, which is calculated (just once) by @code{print-graph}. @smallexample @group (defun Y-axis-element (number full-Y-label-width) "Construct a NUMBERed label element. A numbered element looks like this ` 5 - ', and is padded as needed so all line up with the element for the largest number." @end group @group (let* ((leading-spaces (- full-Y-label-width (length (concat (number-to-string number) Y-axis-tic))))) @end group @group (concat (make-string leading-spaces ? ) (number-to-string number) Y-axis-tic))) @end group @end smallexample The @code{Y-axis-element} function concatenates together the leading spaces, if any; the number, as a string; and the tic mark. To figure out how many leading spaces the label will need, the function subtracts the actual length of the label---the length of the number plus the length of the tic mark---from the desired label width. @findex make-string Blank spaces are inserted using the @code{make-string} function. This function takes two arguments: the first tells it how long the string will be and the second is a symbol for the character to insert, in a special format. The format is a question mark followed by a blank space, like this, @samp{? }. @xref{Character Type, , Character Type, elisp, The GNU Emacs Lisp Reference Manual}, for a description of the syntax for characters. (Of course, you might want to replace the blank space by some other character @dots{} You know what to do.) The @code{number-to-string} function is used in the concatenation expression, to convert the number to a string that is concatenated with the leading spaces and the tic mark. @node Y-axis-column @appendixsubsec Create a Y Axis Column The preceding functions provide all the tools needed to construct a function that generates a list of numbered and blank strings to insert as the label for the vertical axis: @findex Y-axis-column @smallexample @group (defun Y-axis-column (height width-of-label) "Construct list of Y axis labels and blank strings. For HEIGHT of line above base and WIDTH-OF-LABEL." (let (Y-axis) @group @end group (while (> height 1) (if (zerop (% height Y-axis-label-spacing)) ;; @r{Insert label.} (setq Y-axis (cons (Y-axis-element height width-of-label) Y-axis)) @group @end group ;; @r{Else, insert blanks.} (setq Y-axis (cons (make-string width-of-label ? ) Y-axis))) (setq height (1- height))) ;; @r{Insert base line.} (setq Y-axis (cons (Y-axis-element 1 width-of-label) Y-axis)) (nreverse Y-axis))) @end group @end smallexample In this function, we start with the value of @code{height} and repetitively subtract one from its value. After each subtraction, we test to see whether the value is an integral multiple of the @code{Y-axis-label-spacing}. If it is, we construct a numbered label using the @code{Y-axis-element} function; if not, we construct a blank label using the @code{make-string} function. The base line consists of the number one followed by a tic mark. @need 2000 @node print-Y-axis Penultimate @appendixsubsec The Not Quite Final Version of @code{print-Y-axis} The list constructed by the @code{Y-axis-column} function is passed to the @code{print-Y-axis} function, which inserts the list as a column. @findex print-Y-axis @smallexample @group (defun print-Y-axis (height full-Y-label-width) "Insert Y axis using HEIGHT and FULL-Y-LABEL-WIDTH. Height must be the maximum height of the graph. Full width is the width of the highest label element." ;; Value of height and full-Y-label-width ;; are passed by print-graph. @end group @group (let ((start (point))) (insert-rectangle (Y-axis-column height full-Y-label-width)) ;; @r{Place point ready for inserting graph.} (goto-char start) ;; @r{Move point forward by value of} full-Y-label-width (forward-char full-Y-label-width))) @end group @end smallexample The @code{print-Y-axis} uses the @code{insert-rectangle} function to insert the Y axis labels created by the @code{Y-axis-column} function. In addition, it places point at the correct position for printing the body of the graph. You can test @code{print-Y-axis}: @enumerate @item Install @smallexample @group Y-axis-label-spacing Y-axis-tic Y-axis-element Y-axis-column print-Y-axis @end group @end smallexample @item Copy the following expression: @smallexample (print-Y-axis 12 5) @end smallexample @item Switch to the @file{*scratch*} buffer and place the cursor where you want the axis labels to start. @item Type @kbd{M-:} (@code{eval-expression}). @item Yank the @code{graph-body-print} expression into the minibuffer with @kbd{C-y} (@code{yank)}. @item Press @key{RET} to evaluate the expression. @end enumerate Emacs will print labels vertically, the top one being @w{@samp{10 -@w{ }}}. (The @code{print-graph} function will pass the value of @code{height-of-top-line}, which in this case will end up as 15, thereby getting rid of what might appear as a bug.) @need 2000 @node print-X-axis @appendixsec The @code{print-X-axis} Function @cindex Axis, print horizontal @cindex X axis printing @cindex Print horizontal axis @cindex Horizontal axis printing X axis labels are much like Y axis labels, except that the ticks are on a line above the numbers. Labels should look like this: @smallexample @group | | | | 1 5 10 15 @end group @end smallexample The first tic is under the first column of the graph and is preceded by several blank spaces. These spaces provide room in rows above for the Y axis labels. The second, third, fourth, and subsequent ticks are all spaced equally, according to the value of @code{X-axis-label-spacing}. The second row of the X axis consists of numbers, preceded by several blank spaces and also separated according to the value of the variable @code{X-axis-label-spacing}. The value of the variable @code{X-axis-label-spacing} should itself be measured in units of @code{symbol-width}, since you may want to change the width of the symbols that you are using to print the body of the graph without changing the ways the graph is labeled. @menu * Similarities differences:: Much like @code{print-Y-axis}, but not exactly. * X Axis Tic Marks:: Create tic marks for the horizontal axis. @end menu @ifnottex @node Similarities differences @unnumberedsubsec Similarities and differences @end ifnottex The @code{print-X-axis} function is constructed in more or less the same fashion as the @code{print-Y-axis} function except that it has two lines: the line of tic marks and the numbers. We will write a separate function to print each line and then combine them within the @code{print-X-axis} function. This is a three step process: @enumerate @item Write a function to print the X axis tic marks, @code{print-X-axis-tic-line}. @item Write a function to print the X numbers, @code{print-X-axis-numbered-line}. @item Write a function to print both lines, the @code{print-X-axis} function, using @code{print-X-axis-tic-line} and @code{print-X-axis-numbered-line}. @end enumerate @node X Axis Tic Marks @appendixsubsec X Axis Tic Marks The first function should print the X axis tic marks. We must specify the tic marks themselves and their spacing: @smallexample @group (defvar X-axis-label-spacing (if (boundp 'graph-blank) (* 5 (length graph-blank)) 5) "Number of units from one X axis label to next.") @end group @end smallexample @noindent (Note that the value of @code{graph-blank} is set by another @code{defvar}. The @code{boundp} predicate checks whether it has already been set; @code{boundp} returns @code{nil} if it has not. If @code{graph-blank} were unbound and we did not use this conditional construction, in a recent GNU Emacs, we would enter the debugger and see an error message saying @samp{@w{Debugger entered--Lisp error:} @w{(void-variable graph-blank)}}.) @need 1200 Here is the @code{defvar} for @code{X-axis-tic-symbol}: @smallexample @group (defvar X-axis-tic-symbol "|" "String to insert to point to a column in X axis.") @end group @end smallexample @need 1250 The goal is to make a line that looks like this: @smallexample | | | | @end smallexample The first tic is indented so that it is under the first column, which is indented to provide space for the Y axis labels. A tic element consists of the blank spaces that stretch from one tic to the next plus a tic symbol. The number of blanks is determined by the width of the tic symbol and the @code{X-axis-label-spacing}. @need 1250 The code looks like this: @smallexample @group ;;; X-axis-tic-element @dots{} (concat (make-string ;; @r{Make a string of blanks.} (- (* symbol-width X-axis-label-spacing) (length X-axis-tic-symbol)) ? ) ;; @r{Concatenate blanks with tic symbol.} X-axis-tic-symbol) @dots{} @end group @end smallexample Next, we determine how many blanks are needed to indent the first tic mark to the first column of the graph. This uses the value of @code{full-Y-label-width} passed it by the @code{print-graph} function. @need 1250 The code to make @code{X-axis-leading-spaces} looks like this: @smallexample @group ;; X-axis-leading-spaces @dots{} (make-string full-Y-label-width ? ) @dots{} @end group @end smallexample We also need to determine the length of the horizontal axis, which is the length of the numbers list, and the number of ticks in the horizontal axis: @smallexample @group ;; X-length @dots{} (length numbers-list) @end group @group ;; tic-width @dots{} (* symbol-width X-axis-label-spacing) @end group @group ;; number-of-X-ticks (if (zerop (% (X-length tic-width))) (/ (X-length tic-width)) (1+ (/ (X-length tic-width)))) @end group @end smallexample @need 1250 All this leads us directly to the function for printing the X axis tic line: @findex print-X-axis-tic-line @smallexample @group (defun print-X-axis-tic-line (number-of-X-tics X-axis-leading-spaces X-axis-tic-element) "Print ticks for X axis." (insert X-axis-leading-spaces) (insert X-axis-tic-symbol) ; @r{Under first column.} @end group @group ;; @r{Insert second tic in the right spot.} (insert (concat (make-string (- (* symbol-width X-axis-label-spacing) ;; @r{Insert white space up to second tic symbol.} (* 2 (length X-axis-tic-symbol))) ? ) X-axis-tic-symbol)) @end group @group ;; @r{Insert remaining ticks.} (while (> number-of-X-tics 1) (insert X-axis-tic-element) (setq number-of-X-tics (1- number-of-X-tics)))) @end group @end smallexample The line of numbers is equally straightforward: @need 1250 First, we create a numbered element with blank spaces before each number: @findex X-axis-element @smallexample @group (defun X-axis-element (number) "Construct a numbered X axis element." (let ((leading-spaces (- (* symbol-width X-axis-label-spacing) (length (number-to-string number))))) (concat (make-string leading-spaces ? ) (number-to-string number)))) @end group @end smallexample Next, we create the function to print the numbered line, starting with the number 1 under the first column: @findex print-X-axis-numbered-line @smallexample @group (defun print-X-axis-numbered-line (number-of-X-tics X-axis-leading-spaces) "Print line of X-axis numbers" (let ((number X-axis-label-spacing)) (insert X-axis-leading-spaces) (insert "1") @end group @group (insert (concat (make-string ;; @r{Insert white space up to next number.} (- (* symbol-width X-axis-label-spacing) 2) ? ) (number-to-string number))) @end group @group ;; @r{Insert remaining numbers.} (setq number (+ number X-axis-label-spacing)) (while (> number-of-X-tics 1) (insert (X-axis-element number)) (setq number (+ number X-axis-label-spacing)) (setq number-of-X-tics (1- number-of-X-tics))))) @end group @end smallexample Finally, we need to write the @code{print-X-axis} that uses @code{print-X-axis-tic-line} and @code{print-X-axis-numbered-line}. The function must determine the local values of the variables used by both @code{print-X-axis-tic-line} and @code{print-X-axis-numbered-line}, and then it must call them. Also, it must print the carriage return that separates the two lines. The function consists of a varlist that specifies five local variables, and calls to each of the two line printing functions: @findex print-X-axis @smallexample @group (defun print-X-axis (numbers-list) "Print X axis labels to length of NUMBERS-LIST." (let* ((leading-spaces (make-string full-Y-label-width ? )) @end group @group ;; symbol-width @r{is provided by} graph-body-print (tic-width (* symbol-width X-axis-label-spacing)) (X-length (length numbers-list)) @end group @group (X-tic (concat (make-string @end group @group ;; @r{Make a string of blanks.} (- (* symbol-width X-axis-label-spacing) (length X-axis-tic-symbol)) ? ) @end group @group ;; @r{Concatenate blanks with tic symbol.} X-axis-tic-symbol)) @end group @group (tic-number (if (zerop (% X-length tic-width)) (/ X-length tic-width) (1+ (/ X-length tic-width))))) @end group @group (print-X-axis-tic-line tic-number leading-spaces X-tic) (insert "\n") (print-X-axis-numbered-line tic-number leading-spaces))) @end group @end smallexample @need 1250 You can test @code{print-X-axis}: @enumerate @item Install @code{X-axis-tic-symbol}, @code{X-axis-label-spacing}, @code{print-X-axis-tic-line}, as well as @code{X-axis-element}, @code{print-X-axis-numbered-line}, and @code{print-X-axis}. @item Copy the following expression: @smallexample @group (progn (let ((full-Y-label-width 5) (symbol-width 1)) (print-X-axis '(1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16)))) @end group @end smallexample @item Switch to the @file{*scratch*} buffer and place the cursor where you want the axis labels to start. @item Type @kbd{M-:} (@code{eval-expression}). @item Yank the test expression into the minibuffer with @kbd{C-y} (@code{yank)}. @item Press @key{RET} to evaluate the expression. @end enumerate @need 1250 Emacs will print the horizontal axis like this: @sp 1 @smallexample @group | | | | | 1 5 10 15 20 @end group @end smallexample @node Print Whole Graph @appendixsec Printing the Whole Graph @cindex Printing the whole graph @cindex Whole graph printing @cindex Graph, printing all Now we are nearly ready to print the whole graph. The function to print the graph with the proper labels follows the outline we created earlier (@pxref{Full Graph, , A Graph with Labeled Axes}), but with additions. @need 1250 Here is the outline: @smallexample @group (defun print-graph (numbers-list) "@var{documentation}@dots{}" (let ((height @dots{} @dots{})) @end group @group (print-Y-axis height @dots{} ) (graph-body-print numbers-list) (print-X-axis @dots{} ))) @end group @end smallexample @menu * The final version:: A few changes. * Test print-graph:: Run a short test. * Graphing words in defuns:: Executing the final code. * lambda:: How to write an anonymous function. * mapcar:: Apply a function to elements of a list. * Another Bug:: Yet another bug @dots{} most insidious. * Final printed graph:: The graph itself! @end menu @ifnottex @node The final version @unnumberedsubsec Changes for the Final Version @end ifnottex The final version is different from what we planned in two ways: first, it contains additional values calculated once in the varlist; second, it carries an option to specify the labels' increment per row. This latter feature turns out to be essential; otherwise, a graph may have more rows than fit on a display or on a sheet of paper. @need 1500 This new feature requires a change to the @code{Y-axis-column} function, to add @code{vertical-step} to it. The function looks like this: @findex Y-axis-column @r{Final version.} @smallexample @group ;;; @r{Final version.} (defun Y-axis-column (height width-of-label &optional vertical-step) "Construct list of labels for Y axis. HEIGHT is maximum height of graph. WIDTH-OF-LABEL is maximum width of label. VERTICAL-STEP, an option, is a positive integer that specifies how much a Y axis label increments for each line. For example, a step of 5 means that each line is five units of the graph." @end group @group (let (Y-axis (number-per-line (or vertical-step 1))) (while (> height 1) (if (zerop (% height Y-axis-label-spacing)) @end group @group ;; @r{Insert label.} (setq Y-axis (cons (Y-axis-element (* height number-per-line) width-of-label) Y-axis)) @end group @group ;; @r{Else, insert blanks.} (setq Y-axis (cons (make-string width-of-label ? ) Y-axis))) (setq height (1- height))) @end group @group ;; @r{Insert base line.} (setq Y-axis (cons (Y-axis-element (or vertical-step 1) width-of-label) Y-axis)) (nreverse Y-axis))) @end group @end smallexample The values for the maximum height of graph and the width of a symbol are computed by @code{print-graph} in its @code{let} expression; so @code{graph-body-print} must be changed to accept them. @findex graph-body-print @r{Final version.} @smallexample @group ;;; @r{Final version.} (defun graph-body-print (numbers-list height symbol-width) "Print a bar graph of the NUMBERS-LIST. The numbers-list consists of the Y-axis values. HEIGHT is maximum height of graph. SYMBOL-WIDTH is number of each column." @end group @group (let (from-position) (while numbers-list (setq from-position (point)) (insert-rectangle (column-of-graph height (car numbers-list))) (goto-char from-position) (forward-char symbol-width) @end group @group ;; @r{Draw graph column by column.} (sit-for 0) (setq numbers-list (cdr numbers-list))) ;; @r{Place point for X axis labels.} (forward-line height) (insert "\n"))) @end group @end smallexample @need 1250 Finally, the code for the @code{print-graph} function: @findex print-graph @r{Final version.} @smallexample @group ;;; @r{Final version.} (defun print-graph (numbers-list &optional vertical-step) "Print labeled bar graph of the NUMBERS-LIST. The numbers-list consists of the Y-axis values. @end group @group Optionally, VERTICAL-STEP, a positive integer, specifies how much a Y axis label increments for each line. For example, a step of 5 means that each row is five units." @end group @group (let* ((symbol-width (length graph-blank)) ;; @code{height} @r{is both the largest number} ;; @r{and the number with the most digits.} (height (apply 'max numbers-list)) @end group @group (height-of-top-line (if (zerop (% height Y-axis-label-spacing)) height ;; @r{else} (* (1+ (/ height Y-axis-label-spacing)) Y-axis-label-spacing))) @end group @group (vertical-step (or vertical-step 1)) (full-Y-label-width (length @end group @group (concat (number-to-string (* height-of-top-line vertical-step)) Y-axis-tic)))) @end group @group (print-Y-axis height-of-top-line full-Y-label-width vertical-step) @end group @group (graph-body-print numbers-list height-of-top-line symbol-width) (print-X-axis numbers-list))) @end group @end smallexample @node Test print-graph @appendixsubsec Testing @code{print-graph} @need 1250 We can test the @code{print-graph} function with a short list of numbers: @enumerate @item Install the final versions of @code{Y-axis-column}, @code{graph-body-print}, and @code{print-graph} (in addition to the rest of the code.) @item Copy the following expression: @smallexample (print-graph '(3 2 5 6 7 5 3 4 6 4 3 2 1)) @end smallexample @item Switch to the @file{*scratch*} buffer and place the cursor where you want the axis labels to start. @item Type @kbd{M-:} (@code{eval-expression}). @item Yank the test expression into the minibuffer with @kbd{C-y} (@code{yank)}. @item Press @key{RET} to evaluate the expression. @end enumerate @need 1250 Emacs will print a graph that looks like this: @smallexample @group 10 - * ** * 5 - **** * **** *** * ********* ************ 1 - ************* | | | | 1 5 10 15 @end group @end smallexample @need 1200 On the other hand, if you pass @code{print-graph} a @code{vertical-step} value of 2, by evaluating this expression: @smallexample (print-graph '(3 2 5 6 7 5 3 4 6 4 3 2 1) 2) @end smallexample @need 1250 @noindent The graph looks like this: @smallexample @group 20 - * ** * 10 - **** * **** *** * ********* ************ 2 - ************* | | | | 1 5 10 15 @end group @end smallexample @noindent (A question: is the @samp{2} on the bottom of the vertical axis a bug or a feature? If you think it is a bug, and should be a @samp{1} instead, (or even a @samp{0}), you can modify the sources.) @node Graphing words in defuns @appendixsubsec Graphing Numbers of Words and Symbols Now for the graph for which all this code was written: a graph that shows how many function definitions contain fewer than 10 words and symbols, how many contain between 10 and 19 words and symbols, how many contain between 20 and 29 words and symbols, and so on. This is a multi-step process. First make sure you have loaded all the requisite code. @need 1500 It is a good idea to reset the value of @code{top-of-ranges} in case you have set it to some different value. You can evaluate the following: @smallexample @group (setq top-of-ranges '(10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250 260 270 280 290 300) @end group @end smallexample @noindent Next create a list of the number of words and symbols in each range. @need 1500 @noindent Evaluate the following: @smallexample @group (setq list-for-graph (defuns-per-range (sort (recursive-lengths-list-many-files (directory-files "/usr/local/emacs/lisp" t ".+el$")) '<) top-of-ranges)) @end group @end smallexample @noindent On my old machine, this took about an hour. It looked though 303 Lisp files in my copy of Emacs version 19.23. After all that computing, the @code{list-for-graph} had this value: @smallexample @group (537 1027 955 785 594 483 349 292 224 199 166 120 116 99 90 80 67 48 52 45 41 33 28 26 25 20 12 28 11 13 220) @end group @end smallexample @noindent This means that my copy of Emacs had 537 function definitions with fewer than 10 words or symbols in them, 1,027 function definitions with 10 to 19 words or symbols in them, 955 function definitions with 20 to 29 words or symbols in them, and so on. Clearly, just by looking at this list we can see that most function definitions contain ten to thirty words and symbols. Now for printing. We do @emph{not} want to print a graph that is 1,030 lines high @dots{} Instead, we should print a graph that is fewer than twenty-five lines high. A graph that height can be displayed on almost any monitor, and easily printed on a sheet of paper. This means that each value in @code{list-for-graph} must be reduced to one-fiftieth its present value. Here is a short function to do just that, using two functions we have not yet seen, @code{mapcar} and @code{lambda}. @smallexample @group (defun one-fiftieth (full-range) "Return list, each number one-fiftieth of previous." (mapcar (lambda (arg) (/ arg 50)) full-range)) @end group @end smallexample @node lambda @appendixsubsec A @code{lambda} Expression: Useful Anonymity @cindex Anonymous function @findex lambda @code{lambda} is the symbol for an anonymous function, a function without a name. Every time you use an anonymous function, you need to include its whole body. @need 1250 @noindent Thus, @smallexample (lambda (arg) (/ arg 50)) @end smallexample @noindent is a function that returns the value resulting from dividing whatever is passed to it as @code{arg} by 50. @need 1200 Earlier, for example, we had a function @code{multiply-by-seven}; it multiplied its argument by 7. This function is similar, except it divides its argument by 50; and, it has no name. The anonymous equivalent of @code{multiply-by-seven} is: @smallexample (lambda (number) (* 7 number)) @end smallexample @noindent (@xref{defun, , The @code{defun} Macro}.) @need 1250 @noindent If we want to multiply 3 by 7, we can write: @c clear print-postscript-figures @c lambda example diagram #1 @ifnottex @smallexample @group (multiply-by-seven 3) \_______________/ ^ | | function argument @end group @end smallexample @end ifnottex @ifset print-postscript-figures @sp 1 @tex @center @image{lambda-1} @end tex @sp 1 @end ifset @ifclear print-postscript-figures @iftex @smallexample @group (multiply-by-seven 3) \_______________/ ^ | | function argument @end group @end smallexample @end iftex @end ifclear @noindent This expression returns 21. @need 1250 @noindent Similarly, we can write: @c lambda example diagram #2 @ifnottex @smallexample @group ((lambda (number) (* 7 number)) 3) \____________________________/ ^ | | anonymous function argument @end group @end smallexample @end ifnottex @ifset print-postscript-figures @sp 1 @tex @center @image{lambda-2} @end tex @sp 1 @end ifset @ifclear print-postscript-figures @iftex @smallexample @group ((lambda (number) (* 7 number)) 3) \____________________________/ ^ | | anonymous function argument @end group @end smallexample @end iftex @end ifclear @need 1250 @noindent If we want to divide 100 by 50, we can write: @c lambda example diagram #3 @ifnottex @smallexample @group ((lambda (arg) (/ arg 50)) 100) \______________________/ \_/ | | anonymous function argument @end group @end smallexample @end ifnottex @ifset print-postscript-figures @sp 1 @tex @center @image{lambda-3} @end tex @sp 1 @end ifset @ifclear print-postscript-figures @iftex @smallexample @group ((lambda (arg) (/ arg 50)) 100) \______________________/ \_/ | | anonymous function argument @end group @end smallexample @end iftex @end ifclear @noindent This expression returns 2. The 100 is passed to the function, which divides that number by 50. @xref{Lambda Expressions, , Lambda Expressions, elisp, The GNU Emacs Lisp Reference Manual}, for more about @code{lambda}. Lisp and lambda expressions derive from the Lambda Calculus. @node mapcar @appendixsubsec The @code{mapcar} Function @findex mapcar @code{mapcar} is a function that calls its first argument with each element of its second argument, in turn. The second argument must be a sequence. The @samp{map} part of the name comes from the mathematical phrase, ``mapping over a domain'', meaning to apply a function to each of the elements in a domain. The mathematical phrase is based on the metaphor of a surveyor walking, one step at a time, over an area he is mapping. And @samp{car}, of course, comes from the Lisp notion of the first of a list. @need 1250 @noindent For example, @smallexample @group (mapcar '1+ '(2 4 6)) @result{} (3 5 7) @end group @end smallexample @noindent The function @code{1+} which adds one to its argument, is executed on @emph{each} element of the list, and a new list is returned. Contrast this with @code{apply}, which applies its first argument to all the remaining. (@xref{Readying a Graph, , Readying a Graph}, for an explanation of @code{apply}.) @need 1250 In the definition of @code{one-fiftieth}, the first argument is the anonymous function: @smallexample (lambda (arg) (/ arg 50)) @end smallexample @noindent and the second argument is @code{full-range}, which will be bound to @code{list-for-graph}. @need 1250 The whole expression looks like this: @smallexample (mapcar (lambda (arg) (/ arg 50)) full-range)) @end smallexample @xref{Mapping Functions, , Mapping Functions, elisp, The GNU Emacs Lisp Reference Manual}, for more about @code{mapcar}. Using the @code{one-fiftieth} function, we can generate a list in which each element is one-fiftieth the size of the corresponding element in @code{list-for-graph}. @smallexample @group (setq fiftieth-list-for-graph (one-fiftieth list-for-graph)) @end group @end smallexample @need 1250 The resulting list looks like this: @smallexample @group (10 20 19 15 11 9 6 5 4 3 3 2 2 1 1 1 1 0 1 0 0 0 0 0 0 0 0 0 0 0 4) @end group @end smallexample @noindent This, we are almost ready to print! (We also notice the loss of information: many of the higher ranges are 0, meaning that fewer than 50 defuns had that many words or symbols---but not necessarily meaning that none had that many words or symbols.) @node Another Bug @appendixsubsec Another Bug @dots{} Most Insidious @cindex Bug, most insidious type @cindex Insidious type of bug I said ``almost ready to print''! Of course, there is a bug in the @code{print-graph} function @dots{} It has a @code{vertical-step} option, but not a @code{horizontal-step} option. The @code{top-of-range} scale goes from 10 to 300 by tens. But the @code{print-graph} function will print only by ones. This is a classic example of what some consider the most insidious type of bug, the bug of omission. This is not the kind of bug you can find by studying the code, for it is not in the code; it is an omitted feature. Your best actions are to try your program early and often; and try to arrange, as much as you can, to write code that is easy to understand and easy to change. Try to be aware, whenever you can, that whatever you have written, @emph{will} be rewritten, if not soon, eventually. A hard maxim to follow. It is the @code{print-X-axis-numbered-line} function that needs the work; and then the @code{print-X-axis} and the @code{print-graph} functions need to be adapted. Not much needs to be done; there is one nicety: the numbers ought to line up under the tic marks. This takes a little thought. @need 1250 Here is the corrected @code{print-X-axis-numbered-line}: @smallexample @group (defun print-X-axis-numbered-line (number-of-X-tics X-axis-leading-spaces &optional horizontal-step) "Print line of X-axis numbers" (let ((number X-axis-label-spacing) (horizontal-step (or horizontal-step 1))) @end group @group (insert X-axis-leading-spaces) ;; @r{Delete extra leading spaces.} (delete-char (- (1- (length (number-to-string horizontal-step))))) (insert (concat (make-string @end group @group ;; @r{Insert white space.} (- (* symbol-width X-axis-label-spacing) (1- (length (number-to-string horizontal-step))) 2) ? ) (number-to-string (* number horizontal-step)))) @end group @group ;; @r{Insert remaining numbers.} (setq number (+ number X-axis-label-spacing)) (while (> number-of-X-tics 1) (insert (X-axis-element (* number horizontal-step))) (setq number (+ number X-axis-label-spacing)) (setq number-of-X-tics (1- number-of-X-tics))))) @end group @end smallexample @need 1500 If you are reading this in Info, you can see the new versions of @code{print-X-axis} @code{print-graph} and evaluate them. If you are reading this in a printed book, you can see the changed lines here (the full text is too much to print). @iftex @smallexample @group (defun print-X-axis (numbers-list horizontal-step) @dots{} (print-X-axis-numbered-line tic-number leading-spaces horizontal-step)) @end group @end smallexample @smallexample @group (defun print-graph (numbers-list &optional vertical-step horizontal-step) @dots{} (print-X-axis numbers-list horizontal-step)) @end group @end smallexample @end iftex @ifnottex @smallexample @group (defun print-X-axis (numbers-list horizontal-step) "Print X axis labels to length of NUMBERS-LIST. Optionally, HORIZONTAL-STEP, a positive integer, specifies how much an X axis label increments for each column." @end group @group ;; Value of symbol-width and full-Y-label-width ;; are passed by print-graph. (let* ((leading-spaces (make-string full-Y-label-width ? )) ;; symbol-width @r{is provided by} graph-body-print (tic-width (* symbol-width X-axis-label-spacing)) (X-length (length numbers-list)) @end group @group (X-tic (concat (make-string ;; @r{Make a string of blanks.} (- (* symbol-width X-axis-label-spacing) (length X-axis-tic-symbol)) ? ) @end group @group ;; @r{Concatenate blanks with tic symbol.} X-axis-tic-symbol)) (tic-number (if (zerop (% X-length tic-width)) (/ X-length tic-width) (1+ (/ X-length tic-width))))) @end group @group (print-X-axis-tic-line tic-number leading-spaces X-tic) (insert "\n") (print-X-axis-numbered-line tic-number leading-spaces horizontal-step))) @end group @end smallexample @smallexample @group (defun print-graph (numbers-list &optional vertical-step horizontal-step) "Print labeled bar graph of the NUMBERS-LIST. The numbers-list consists of the Y-axis values. @end group @group Optionally, VERTICAL-STEP, a positive integer, specifies how much a Y axis label increments for each line. For example, a step of 5 means that each row is five units. @end group @group Optionally, HORIZONTAL-STEP, a positive integer, specifies how much an X axis label increments for each column." (let* ((symbol-width (length graph-blank)) ;; @code{height} @r{is both the largest number} ;; @r{and the number with the most digits.} (height (apply 'max numbers-list)) @end group @group (height-of-top-line (if (zerop (% height Y-axis-label-spacing)) height ;; @r{else} (* (1+ (/ height Y-axis-label-spacing)) Y-axis-label-spacing))) @end group @group (vertical-step (or vertical-step 1)) (full-Y-label-width (length (concat (number-to-string (* height-of-top-line vertical-step)) Y-axis-tic)))) @end group @group (print-Y-axis height-of-top-line full-Y-label-width vertical-step) (graph-body-print numbers-list height-of-top-line symbol-width) (print-X-axis numbers-list horizontal-step))) @end group @end smallexample @end ifnottex @c qqq @ignore Graphing Definitions Re-listed @need 1250 Here are all the graphing definitions in their final form: @smallexample @group (defvar top-of-ranges '(10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250) "List specifying ranges for `defuns-per-range'.") @end group @group (defvar graph-symbol "*" "String used as symbol in graph, usually an asterisk.") @end group @group (defvar graph-blank " " "String used as blank in graph, usually a blank space. graph-blank must be the same number of columns wide as graph-symbol.") @end group @group (defvar Y-axis-tic " - " "String that follows number in a Y axis label.") @end group @group (defvar Y-axis-label-spacing 5 "Number of lines from one Y axis label to next.") @end group @group (defvar X-axis-tic-symbol "|" "String to insert to point to a column in X axis.") @end group @group (defvar X-axis-label-spacing (if (boundp 'graph-blank) (* 5 (length graph-blank)) 5) "Number of units from one X axis label to next.") @end group @end smallexample @smallexample @group (defun count-words-in-defun () "Return the number of words and symbols in a defun." (beginning-of-defun) (let ((count 0) (end (save-excursion (end-of-defun) (point)))) @end group @group (while (and (< (point) end) (re-search-forward "\\(\\w\\|\\s_\\)+[^ \t\n]*[ \t\n]*" end t)) (setq count (1+ count))) count)) @end group @end smallexample @smallexample @group (defun lengths-list-file (filename) "Return list of definitions' lengths within FILE. The returned list is a list of numbers. Each number is the number of words or symbols in one function definition." @end group @group (message "Working on `%s' ... " filename) (save-excursion (let ((buffer (find-file-noselect filename)) (lengths-list)) (set-buffer buffer) (setq buffer-read-only t) (widen) (goto-char (point-min)) @end group @group (while (re-search-forward "^(defun" nil t) (setq lengths-list (cons (count-words-in-defun) lengths-list))) (kill-buffer buffer) lengths-list))) @end group @end smallexample @smallexample @group (defun lengths-list-many-files (list-of-files) "Return list of lengths of defuns in LIST-OF-FILES." (let (lengths-list) ;;; @r{true-or-false-test} (while list-of-files (setq lengths-list (append lengths-list @end group @group ;;; @r{Generate a lengths' list.} (lengths-list-file (expand-file-name (car list-of-files))))) ;;; @r{Make files' list shorter.} (setq list-of-files (cdr list-of-files))) ;;; @r{Return final value of lengths' list.} lengths-list)) @end group @end smallexample @smallexample @group (defun defuns-per-range (sorted-lengths top-of-ranges) "SORTED-LENGTHS defuns in each TOP-OF-RANGES range." (let ((top-of-range (car top-of-ranges)) (number-within-range 0) defuns-per-range-list) @end group @group ;; @r{Outer loop.} (while top-of-ranges ;; @r{Inner loop.} (while (and ;; @r{Need number for numeric test.} (car sorted-lengths) (< (car sorted-lengths) top-of-range)) ;; @r{Count number of definitions within current range.} (setq number-within-range (1+ number-within-range)) (setq sorted-lengths (cdr sorted-lengths))) @end group @group ;; @r{Exit inner loop but remain within outer loop.} (setq defuns-per-range-list (cons number-within-range defuns-per-range-list)) (setq number-within-range 0) ; @r{Reset count to zero.} ;; @r{Move to next range.} (setq top-of-ranges (cdr top-of-ranges)) ;; @r{Specify next top of range value.} (setq top-of-range (car top-of-ranges))) @end group @group ;; @r{Exit outer loop and count the number of defuns larger than} ;; @r{ the largest top-of-range value.} (setq defuns-per-range-list (cons (length sorted-lengths) defuns-per-range-list)) ;; @r{Return a list of the number of definitions within each range,} ;; @r{ smallest to largest.} (nreverse defuns-per-range-list))) @end group @end smallexample @smallexample @group (defun column-of-graph (max-graph-height actual-height) "Return list of MAX-GRAPH-HEIGHT strings; ACTUAL-HEIGHT are graph-symbols. The graph-symbols are contiguous entries at the end of the list. The list will be inserted as one column of a graph. The strings are either graph-blank or graph-symbol." @end group @group (let ((insert-list nil) (number-of-top-blanks (- max-graph-height actual-height))) ;; @r{Fill in @code{graph-symbols}.} (while (> actual-height 0) (setq insert-list (cons graph-symbol insert-list)) (setq actual-height (1- actual-height))) @end group @group ;; @r{Fill in @code{graph-blanks}.} (while (> number-of-top-blanks 0) (setq insert-list (cons graph-blank insert-list)) (setq number-of-top-blanks (1- number-of-top-blanks))) ;; @r{Return whole list.} insert-list)) @end group @end smallexample @smallexample @group (defun Y-axis-element (number full-Y-label-width) "Construct a NUMBERed label element. A numbered element looks like this ` 5 - ', and is padded as needed so all line up with the element for the largest number." @end group @group (let* ((leading-spaces (- full-Y-label-width (length (concat (number-to-string number) Y-axis-tic))))) @end group @group (concat (make-string leading-spaces ? ) (number-to-string number) Y-axis-tic))) @end group @end smallexample @smallexample @group (defun print-Y-axis (height full-Y-label-width &optional vertical-step) "Insert Y axis by HEIGHT and FULL-Y-LABEL-WIDTH. Height must be the maximum height of the graph. Full width is the width of the highest label element. Optionally, print according to VERTICAL-STEP." @end group @group ;; Value of height and full-Y-label-width ;; are passed by 'print-graph'. (let ((start (point))) (insert-rectangle (Y-axis-column height full-Y-label-width vertical-step)) @end group @group ;; @r{Place point ready for inserting graph.} (goto-char start) ;; @r{Move point forward by value of} full-Y-label-width (forward-char full-Y-label-width))) @end group @end smallexample @smallexample @group (defun print-X-axis-tic-line (number-of-X-tics X-axis-leading-spaces X-axis-tic-element) "Print ticks for X axis." (insert X-axis-leading-spaces) (insert X-axis-tic-symbol) ; @r{Under first column.} @end group @group ;; @r{Insert second tic in the right spot.} (insert (concat (make-string (- (* symbol-width X-axis-label-spacing) ;; @r{Insert white space up to second tic symbol.} (* 2 (length X-axis-tic-symbol))) ? ) X-axis-tic-symbol)) @end group @group ;; @r{Insert remaining ticks.} (while (> number-of-X-tics 1) (insert X-axis-tic-element) (setq number-of-X-tics (1- number-of-X-tics)))) @end group @end smallexample @smallexample @group (defun X-axis-element (number) "Construct a numbered X axis element." (let ((leading-spaces (- (* symbol-width X-axis-label-spacing) (length (number-to-string number))))) (concat (make-string leading-spaces ? ) (number-to-string number)))) @end group @end smallexample @smallexample @group (defun graph-body-print (numbers-list height symbol-width) "Print a bar graph of the NUMBERS-LIST. The numbers-list consists of the Y-axis values. HEIGHT is maximum height of graph. SYMBOL-WIDTH is number of each column." @end group @group (let (from-position) (while numbers-list (setq from-position (point)) (insert-rectangle (column-of-graph height (car numbers-list))) (goto-char from-position) (forward-char symbol-width) @end group @group ;; @r{Draw graph column by column.} (sit-for 0) (setq numbers-list (cdr numbers-list))) ;; @r{Place point for X axis labels.} (forward-line height) (insert "\n"))) @end group @end smallexample @smallexample @group (defun Y-axis-column (height width-of-label &optional vertical-step) "Construct list of labels for Y axis. HEIGHT is maximum height of graph. WIDTH-OF-LABEL is maximum width of label. @end group @group VERTICAL-STEP, an option, is a positive integer that specifies how much a Y axis label increments for each line. For example, a step of 5 means that each line is five units of the graph." (let (Y-axis (number-per-line (or vertical-step 1))) @end group @group (while (> height 1) (if (zerop (% height Y-axis-label-spacing)) ;; @r{Insert label.} (setq Y-axis (cons (Y-axis-element (* height number-per-line) width-of-label) Y-axis)) @end group @group ;; @r{Else, insert blanks.} (setq Y-axis (cons (make-string width-of-label ? ) Y-axis))) (setq height (1- height))) @end group @group ;; @r{Insert base line.} (setq Y-axis (cons (Y-axis-element (or vertical-step 1) width-of-label) Y-axis)) (nreverse Y-axis))) @end group @end smallexample @smallexample @group (defun print-X-axis-numbered-line (number-of-X-tics X-axis-leading-spaces &optional horizontal-step) "Print line of X-axis numbers" (let ((number X-axis-label-spacing) (horizontal-step (or horizontal-step 1))) @end group @group (insert X-axis-leading-spaces) ;; line up number (delete-char (- (1- (length (number-to-string horizontal-step))))) (insert (concat (make-string ;; @r{Insert white space up to next number.} (- (* symbol-width X-axis-label-spacing) (1- (length (number-to-string horizontal-step))) 2) ? ) (number-to-string (* number horizontal-step)))) @end group @group ;; @r{Insert remaining numbers.} (setq number (+ number X-axis-label-spacing)) (while (> number-of-X-tics 1) (insert (X-axis-element (* number horizontal-step))) (setq number (+ number X-axis-label-spacing)) (setq number-of-X-tics (1- number-of-X-tics))))) @end group @end smallexample @smallexample @group (defun print-X-axis (numbers-list horizontal-step) "Print X axis labels to length of NUMBERS-LIST. Optionally, HORIZONTAL-STEP, a positive integer, specifies how much an X axis label increments for each column." @end group @group ;; Value of symbol-width and full-Y-label-width ;; are passed by 'print-graph'. (let* ((leading-spaces (make-string full-Y-label-width ? )) ;; symbol-width @r{is provided by} graph-body-print (tic-width (* symbol-width X-axis-label-spacing)) (X-length (length numbers-list)) @end group @group (X-tic (concat (make-string ;; @r{Make a string of blanks.} (- (* symbol-width X-axis-label-spacing) (length X-axis-tic-symbol)) ? ) @end group @group ;; @r{Concatenate blanks with tic symbol.} X-axis-tic-symbol)) (tic-number (if (zerop (% X-length tic-width)) (/ X-length tic-width) (1+ (/ X-length tic-width))))) @end group @group (print-X-axis-tic-line tic-number leading-spaces X-tic) (insert "\n") (print-X-axis-numbered-line tic-number leading-spaces horizontal-step))) @end group @end smallexample @smallexample @group (defun one-fiftieth (full-range) "Return list, each number of which is 1/50th previous." (mapcar (lambda (arg) (/ arg 50)) full-range)) @end group @end smallexample @smallexample @group (defun print-graph (numbers-list &optional vertical-step horizontal-step) "Print labeled bar graph of the NUMBERS-LIST. The numbers-list consists of the Y-axis values. @end group @group Optionally, VERTICAL-STEP, a positive integer, specifies how much a Y axis label increments for each line. For example, a step of 5 means that each row is five units. @end group @group Optionally, HORIZONTAL-STEP, a positive integer, specifies how much an X axis label increments for each column." (let* ((symbol-width (length graph-blank)) ;; @code{height} @r{is both the largest number} ;; @r{and the number with the most digits.} (height (apply 'max numbers-list)) @end group @group (height-of-top-line (if (zerop (% height Y-axis-label-spacing)) height ;; @r{else} (* (1+ (/ height Y-axis-label-spacing)) Y-axis-label-spacing))) @end group @group (vertical-step (or vertical-step 1)) (full-Y-label-width (length (concat (number-to-string (* height-of-top-line vertical-step)) Y-axis-tic)))) @end group @group (print-Y-axis height-of-top-line full-Y-label-width vertical-step) (graph-body-print numbers-list height-of-top-line symbol-width) (print-X-axis numbers-list horizontal-step))) @end group @end smallexample @c qqq @end ignore @page @node Final printed graph @appendixsubsec The Printed Graph When made and installed, you can call the @code{print-graph} command like this: @sp 1 @smallexample @group (print-graph fiftieth-list-for-graph 50 10) @end group @end smallexample @sp 1 @noindent Here is the graph: @sp 2 @smallexample @group 1000 - * ** ** ** ** 750 - *** *** *** *** **** 500 - ***** ****** ****** ****** ******* 250 - ******** ********* * *********** * ************* * 50 - ***************** * * | | | | | | | | 10 50 100 150 200 250 300 350 @end group @end smallexample @sp 2 @noindent The largest group of functions contain 10--19 words and symbols each. @node Free Software and Free Manuals @appendix Free Software and Free Manuals @strong{by Richard M. Stallman} @sp 1 The biggest deficiency in free operating systems is not in the software---it is the lack of good free manuals that we can include in these systems. Many of our most important programs do not come with full manuals. Documentation is an essential part of any software package; when an important free software package does not come with a free manual, that is a major gap. We have many such gaps today. Once upon a time, many years ago, I thought I would learn Perl. I got a copy of a free manual, but I found it hard to read. When I asked Perl users about alternatives, they told me that there were better introductory manuals---but those were not free. Why was this? The authors of the good manuals had written them for O'Reilly Associates, which published them with restrictive terms---no copying, no modification, source files not available---which exclude them from the free software community. That wasn't the first time this sort of thing has happened, and (to our community's great loss) it was far from the last. Proprietary manual publishers have enticed a great many authors to restrict their manuals since then. Many times I have heard a GNU user eagerly tell me about a manual that he is writing, with which he expects to help the GNU project---and then had my hopes dashed, as he proceeded to explain that he had signed a contract with a publisher that would restrict it so that we cannot use it. Given that writing good English is a rare skill among programmers, we can ill afford to lose manuals this way. Free documentation, like free software, is a matter of freedom, not price. The problem with these manuals was not that O'Reilly Associates charged a price for printed copies---that in itself is fine. The Free Software Foundation @uref{https://shop.fsf.org, sells printed copies} of free @uref{https://www.gnu.org/doc/doc.html, GNU manuals}, too. But GNU manuals are available in source code form, while these manuals are available only on paper. GNU manuals come with permission to copy and modify; the Perl manuals do not. These restrictions are the problems. The criterion for a free manual is pretty much the same as for free software: it is a matter of giving all users certain freedoms. Redistribution (including commercial redistribution) must be permitted, so that the manual can accompany every copy of the program, on-line or on paper. Permission for modification is crucial too. As a general rule, I don't believe that it is essential for people to have permission to modify all sorts of articles and books. The issues for writings are not necessarily the same as those for software. For example, I don't think you or I are obliged to give permission to modify articles like this one, which describe our actions and our views. But there is a particular reason why the freedom to modify is crucial for documentation for free software. When people exercise their right to modify the software, and add or change its features, if they are conscientious they will change the manual too---so they can provide accurate and usable documentation with the modified program. A manual which forbids programmers to be conscientious and finish the job, or more precisely requires them to write a new manual from scratch if they change the program, does not fill our community's needs. While a blanket prohibition on modification is unacceptable, some kinds of limits on the method of modification pose no problem. For example, requirements to preserve the original author's copyright notice, the distribution terms, or the list of authors, are ok. It is also no problem to require modified versions to include notice that they were modified, even to have entire sections that may not be deleted or changed, as long as these sections deal with nontechnical topics. (Some GNU manuals have them.) These kinds of restrictions are not a problem because, as a practical matter, they don't stop the conscientious programmer from adapting the manual to fit the modified program. In other words, they don't block the free software community from making full use of the manual. However, it must be possible to modify all the technical content of the manual, and then distribute the result in all the usual media, through all the usual channels; otherwise, the restrictions do block the community, the manual is not free, and so we need another manual. Unfortunately, it is often hard to find someone to write another manual when a proprietary manual exists. The obstacle is that many users think that a proprietary manual is good enough---so they don't see the need to write a free manual. They do not see that the free operating system has a gap that needs filling. Why do users think that proprietary manuals are good enough? Some have not considered the issue. I hope this article will do something to change that. Other users consider proprietary manuals acceptable for the same reason so many people consider proprietary software acceptable: they judge in purely practical terms, not using freedom as a criterion. These people are entitled to their opinions, but since those opinions spring from values which do not include freedom, they are no guide for those of us who do value freedom. Please spread the word about this issue. We continue to lose manuals to proprietary publishing. If we spread the word that proprietary manuals are not sufficient, perhaps the next person who wants to help GNU by writing documentation will realize, before it is too late, that he must above all make it free. We can also encourage commercial publishers to sell free, copylefted manuals instead of proprietary ones. One way you can help this is to check the distribution terms of a manual before you buy it, and prefer copylefted manuals to non-copylefted ones. @sp 2 @noindent Note: The Free Software Foundation maintains a page on its Web site that lists free books available from other publishers:@* @uref{https://www.gnu.org/doc/other-free-books.html} @node GNU Free Documentation License @appendix GNU Free Documentation License @cindex FDL, GNU Free Documentation License @include doclicense.texi @node Index @unnumbered Index @ignore MENU ENTRY: NODE NAME. @end ignore @printindex cp @iftex @c Place biographical information on right-hand (verso) page @tex \par\vfill\supereject \ifodd\pageno \global\evenheadline={\hfil} \global\evenfootline={\hfil} \global\oddheadline={\hfil} \global\oddfootline={\hfil} %\page\hbox{}\page \else % \par\vfill\supereject \global\evenheadline={\hfil} \global\evenfootline={\hfil} \global\oddheadline={\hfil} \global\oddfootline={\hfil} %\page\hbox{}%\page %\page\hbox{}%\page \fi @end tex @c page @w{ } @c ================ Biographical information ================ @w{ } @sp 8 @center About the Author @sp 1 @end iftex @ifnottex @node About the Author @unnumbered About the Author @end ifnottex @quotation Robert J. Chassell (1946--2017) started working with GNU Emacs in 1985. He wrote and edited, taught Emacs and Emacs Lisp, and spoke throughout the world on software freedom. Chassell was a founding Director and Treasurer of the Free Software Foundation, Inc. He was co-author of the @cite{Texinfo} manual, and edited more than a dozen other books. He graduated from Cambridge University, in England. He had an abiding interest in social and economic history and flew his own airplane. @uref{https://www.fsf.org/blogs/community/goodbye-to-bob-chassell, "Goodbye to Bob Chassell"} @end quotation @c @page @c @w{ } @c @c @c Prevent page number on blank verso, so eject it first. @c @tex @c \par\vfill\supereject @c @end tex @c @iftex @c @headings off @c @evenheading @thispage @| @| @thistitle @c @oddheading @| @| @thispage @c @end iftex @bye