?\_ patch

?\_ patch

[Edward O'Connor]

[-- Attachment #1: Type: text/plain, Size: 760 bytes --]

Since no one seemed to object and it's a trivial change, here's a
patch to add ?\_ support. Light testing here shows no ill effects.

The patch does the following:

1. Adds support for ?\_ to lread.c,
2. Documents ?\_ in lispref,
3. Augments checkdoc to replace end-of-line `? ' and `?\ ' with `?\_',
   (Note that this change isn't as good as it could be, but I don't
    know anything about checkdoc's internals.)
4. Updates etc/NEWS and the various ChangeLogs (which presumably
   should be changed by whoever checks this in).

I hope this change squeaks in under the paperwork threshold (it's
really very very simple), but if not, I'm perfectly willing to
sign papers.


Ted

-- 
Edward O'Connor
ted@oconnor.cx

Ense petit placidam sub libertate quietem.


[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: Add support for ?\\_ escape character --]
[-- Type: text/x-patch, Size: 6237 bytes --]

Index: etc/NEWS
===================================================================
RCS file: /cvsroot/emacs/emacs/etc/NEWS,v
retrieving revision 1.786
diff -u -r1.786 NEWS
--- etc/NEWS	3 Feb 2003 20:55:30 -0000	1.786
+++ etc/NEWS	5 Feb 2003 19:31:13 -0000
@@ -1150,6 +1150,8 @@
 \f
 * Lisp Changes in Emacs 21.4
 
+** New ?\_ escape character for space.
+
 ** New variables `gc-elapsed' and `gcs-done' provide extra information
 on garbage collection.
 
Index: lisp/ChangeLog
===================================================================
RCS file: /cvsroot/emacs/emacs/lisp/ChangeLog,v
retrieving revision 1.4811
diff -u -r1.4811 ChangeLog
--- lisp/ChangeLog	4 Feb 2003 17:07:42 -0000	1.4811
+++ lisp/ChangeLog	5 Feb 2003 19:31:14 -0000
@@ -1,3 +1,8 @@
+2003-02-05  Edward O'Connor  <ted@oconnor.cx>
+	* emacs-lisp/checkdoc.el (checkdoc-rogue-space-check-engine):
+	Replace end-of-line `? ' and `?\ ' with `?\_' instead of
+	requiring a manual fix.
+
 2003-02-04  Richard M. Stallman  <rms@gnu.org>
 
 	* term.el (term-raw-map): Set it up at load time.
Index: lisp/emacs-lisp/checkdoc.el
===================================================================
RCS file: /cvsroot/emacs/emacs/lisp/emacs-lisp/checkdoc.el,v
retrieving revision 1.34
diff -u -r1.34 checkdoc.el
--- lisp/emacs-lisp/checkdoc.el	4 Feb 2003 12:53:34 -0000	1.34
+++ lisp/emacs-lisp/checkdoc.el	5 Feb 2003 19:31:14 -0000
@@ -2112,21 +2112,17 @@
     ;; If end is nil, it means end of buffer to search anyway
     (or
      ;; Check for an error if `? ' or `?\ ' is used at the end of a line.
-     ;; (It's dangerous)
+     ;; This is bad beause `delete-trailing-whitespace' and similar tools
+     ;; may remove this space.
      (progn
        (goto-char start)
        (while (and (not msg) (re-search-forward "\\?\\\\?[ \t][ \t]*$" end t))
 	 (setq msg
 	       "Don't use `? ' at the end of a line. \
-News agents may remove it"
+Replace with ?\\_?"
 	       s (match-beginning 0) e (match-end 0) f t)
-	 ;; If interactive is passed down, give them a chance to fix things.
-	 (if (and interact (y-or-n-p (concat msg ". Fix? ")))
-	     (progn
-	       (checkdoc-recursive-edit msg)
-	       (setq msg nil)
-	       (goto-char s)
-	       (beginning-of-line)))))
+         ;; Replace `? ' and `?\ ' with `?\_'.
+         (checkdoc-autofix-ask-replace s e msg "?\\_")))
      ;; Check for, and potentially remove whitespace appearing at the
      ;; end of different lines.
      (progn
Index: lispref/ChangeLog
===================================================================
RCS file: /cvsroot/emacs/emacs/lispref/ChangeLog,v
retrieving revision 1.63
diff -u -r1.63 ChangeLog
--- lispref/ChangeLog	31 Jan 2003 15:21:47 -0000	1.63
+++ lispref/ChangeLog	5 Feb 2003 19:31:14 -0000
@@ -1,3 +1,8 @@
+2003-02-05  Edward O'Connor  <ted@oconnor.cx>
+
+	* objects.texi (Character Type): Added documentation for new ?\_
+	escape character.
+
 2003-01-31  Joe Buehler  <jhpb@draco.hekimian.com>
 
 	* os.texi (System Environment): Added cygwin system-type.
Index: lispref/objects.texi
===================================================================
RCS file: /cvsroot/emacs/emacs/lispref/objects.texi,v
retrieving revision 1.32
diff -u -r1.32 objects.texi
--- lispref/objects.texi	4 Feb 2003 14:47:54 -0000	1.32
+++ lispref/objects.texi	5 Feb 2003 19:31:14 -0000
@@ -258,11 +258,13 @@
 
   You can use the same syntax for punctuation characters, but it is
 often a good idea to add a @samp{\} so that the Emacs commands for
-editing Lisp code don't get confused.  For example, @samp{?\ } is the
+editing Lisp code don't get confused.  For example, @samp{?\ } is a
 way to write the space character.  If the character is @samp{\}, you
 @emph{must} use a second @samp{\} to quote it: @samp{?\\}.
 
 @cindex whitespace
+@cindex space
+@cindex @samp{\_}
 @cindex bell character
 @cindex @samp{\a}
 @cindex backspace
@@ -279,12 +281,14 @@
 @cindex @samp{\r}
 @cindex escape
 @cindex @samp{\e}
-  You can express the characters Control-g, backspace, tab, newline,
-vertical tab, formfeed, return, del, and escape as @samp{?\a},
-@samp{?\b}, @samp{?\t}, @samp{?\n}, @samp{?\v}, @samp{?\f},
-@samp{?\r}, @samp{?\d}, and @samp{?\e}, respectively.  Thus,
+  You can express the characters space, Control-g, backspace, tab,
+newline, vertical tab, formfeed, return, del, and escape as
+@samp{?\_}, @samp{?\a}, @samp{?\b}, @samp{?\t}, @samp{?\n},
+@samp{?\v}, @samp{?\f}, @samp{?\r}, @samp{?\d}, and @samp{?\e},
+respectively. Thus,
 
 @example
+?\_ @result{} 32                ; @r{space, @key{SPC}}
 ?\a @result{} 7                 ; @r{@kbd{C-g}}
 ?\b @result{} 8                 ; @r{backspace, @key{BS}, @kbd{C-h}}
 ?\t @result{} 9                 ; @r{tab, @key{TAB}, @kbd{C-i}}
@@ -455,8 +459,8 @@
 @samp{()\|;'`"#.,} to avoid confusing the Emacs commands for editing
 Lisp code.  Also add a backslash before whitespace characters such as
 space, tab, newline and formfeed.  However, it is cleaner to use one of
-the easily readable escape sequences, such as @samp{\t}, instead of an
-actual whitespace character such as a tab.
+the easily readable escape sequences, such as @samp{\t} or @samp{\_},
+instead of an actual whitespace character such as a tab or space.
 
 @node Symbol Type
 @subsection Symbol Type
Index: src/ChangeLog
===================================================================
RCS file: /cvsroot/emacs/emacs/src/ChangeLog,v
retrieving revision 1.3081
diff -u -r1.3081 ChangeLog
--- src/ChangeLog	4 Feb 2003 11:52:10 -0000	1.3081
+++ src/ChangeLog	5 Feb 2003 19:31:14 -0000
@@ -1,3 +1,7 @@
+2003-02-05  Edward O'Connor  <ted@oconnor.cx>
+
+	* lread.c (read_escape): Handle new ?\_ escape character.
+
 2003-02-04  Richard M. Stallman  <rms@gnu.org>
 
 	* keyboard.c (echo_now): Update before_command_echo_length.
Index: src/lread.c
===================================================================
RCS file: /cvsroot/emacs/emacs/src/lread.c,v
retrieving revision 1.305
diff -u -r1.305 lread.c
--- src/lread.c	4 Feb 2003 14:03:12 -0000	1.305
+++ src/lread.c	5 Feb 2003 19:31:15 -0000
@@ -1655,6 +1655,8 @@
       return '\v';
     case '\n':
       return -1;
+    case '_':
+      return ' ';
     case ' ':
       if (stringp)
 	return -1;

[-- Attachment #3: Type: text/plain, Size: 142 bytes --]

_______________________________________________
Emacs-devel mailing list
Emacs-devel@gnu.org
http://mail.gnu.org/mailman/listinfo/emacs-devel

Re: ?\_ patch

[Kenichi Handa]

In article <u4cptq6fpzk.fsf@george.floobin.cx>, "Edward O'Connor" <ted@oconnor.cx> writes:
> [1  <text/plain (7bit)>]
> Since no one seemed to object and it's a trivial change, here's a
> patch to add ?\_ support. Light testing here shows no ill effects.

I object to it.

By grepping, I found many files under lisp subdir are using
"?\_" for representing "?_" (e.g. calc/calc-ext.el).  Some
are using ?\X unconditionaly even in the place where ?X
works.

That means there are much more such files not included in
Emacs distribution.  If we change the meaning of ?\_, all of
them stop working.

---
Ken'ichi HANDA
handa@m17n.org

Re: ?\_ patch

[Luc Teirlinck]

Ken'ichi HANDA wrote:

   I object to it.

   By grepping, I found many files under lisp subdir are using
   "?\_" for representing "?_" (e.g. calc/calc-ext.el).  Some
   are using ?\X unconditionaly even in the place where ?X
   works.

   That means there are much more such files not included in
   Emacs distribution.  If we change the meaning of ?\_, all of
   them stop working.

I did not realize people would be doing this, especially since I saw
no recommendation to quote `_' anywhere.

I now agree with you.

This leaves two solutions: use ?\s, which is somewhat confusing, but
perfectly safe since it has been "protected" by an error message for
ages, or change the default value of tabify-regexp to leave ?\ alone.

Of course, the second solution would mean that personal customizations
of tabify-regexp would be dangerous.

I now lean toward the original solution of using ?\s.

Alternatives like using 32,  ?\040 or  ?\x20 do not look attractive.

Sincerely,

Luc.

Re: ?\_ patch

[Luc Teirlinck]

>From my previous message:

   I now lean toward the original solution of using ?\s.

Of course, if one wanted to use it within a string (which one should
allow for consistency), then one might have to worry about whether
"\s-a" would be a valid string, or whether "\s\-a" would be necessary.
Currently, "\s-a" leads to a "Invalid modifier in string" error.
There might be other problems (although I do not see any right now.)

Modifying tabify-regexp is an alternative.  It is not a user option
and one could point out in the documentation string that one should
be careful in modifying it.

Sincerely,

Luc.

Re: ?\_ patch

[Luc Teirlinck]

>From my previous message:

   Of course, if one wanted to use it within a string (which one should
   allow for consistency), then one might have to worry about whether
   "\s-a" would be a valid string, or whether "\s\-a" would be necessary.
   Currently, "\s-a" leads to a "Invalid modifier in string" error.

Requiring "\s\-a" would be better, the alternative would really be too
confusing.

Sincerely,

Luc.

Re: ?\_ patch

[Kenichi Handa]

In article <200302060044.SAA21429@eel.dms.auburn.edu>, Luc Teirlinck <teirllm@dms.auburn.edu> writes:
> I now lean toward the original solution of using ?\s.

I've just found that (eq 's '\s) => t.

But, as (eq 'e '\e) is also t, this is not a problem,
perhaps.

> Alternatives like using 32,  ?\040 or  ?\x20 do not look attractive.

To me the last two are attractive because we only have to
change a manual to promote them instead of "? " or "?\ ". :-)

---
Ken'ichi HANDA
handa@m17n.org

Re: ?\_ patch

[Miles Bader]

storm@cua.dk (Kim F. Storm) writes:
> What about ?\SPC

Only if ?\RET, ?\TAB, etc. work too (otherwise it's just too
inconsistent -- if ?\SPC worked then people would quite naturally expect
things like ?\RET or ?\TAB to work as well).

[I guess there's not a conflict between ?\r and ?\RET because one uses
lowercase and the other uppercase, though it makes me a bit nervous...]

-Miles
-- 
I'd rather be consing.

Re: ?\_ patch

[Kim F. Storm]

Luc Teirlinck <teirllm@dms.auburn.edu> writes:

> >From my previous message:
> 
>    I now lean toward the original solution of using ?\s.
> 
> Of course, if one wanted to use it within a string (which one should
> allow for consistency), then one might have to worry about whether
> "\s-a" would be a valid string, or whether "\s\-a" would be necessary.
> Currently, "\s-a" leads to a "Invalid modifier in string" error.
> There might be other problems (although I do not see any right now.)

What about ?\SPC

Re: ?\_ patch

[Andreas Schwab]

Kenichi Handa <handa@m17n.org> writes:

|> In article <200302060044.SAA21429@eel.dms.auburn.edu>, Luc Teirlinck <teirllm@dms.auburn.edu> writes:
|> > I now lean toward the original solution of using ?\s.
|> 
|> I've just found that (eq 's '\s) => t.
|> 
|> But, as (eq 'e '\e) is also t, this is not a problem,
|> perhaps.

Symbols and characters have quite different quoting rules.  In symbols,
the backslash is always ignored except to make it syntactically different
from a number.

Andreas.

-- 
Andreas Schwab, SuSE Labs, schwab@suse.de
SuSE Linux AG, Deutschherrnstr. 15-19, D-90429 Nürnberg
Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: ?\_ patch

[Juanma Barranquero]

On 06 Feb 2003 12:48:24 +0100, storm@cua.dk (Kim F. Storm) wrote:

> So I suggest that ?\s is a space unless the `s' is immediately followed
> by `-' (e.g. ?\s-a) in which case it has the usual meaning as a super
> modifier.
> 
> In strings, "\s" should also mean a space unless it is following by
> a `-' (as in "\s-a") which should produce an error as it does now.

I like it.


                                                           /L/e/k/t/u

Re: ?\_ patch

[Kim F. Storm]

Miles Bader <miles@lsi.nec.co.jp> writes:

> storm@cua.dk (Kim F. Storm) writes:
> > What about ?\SPC
> 
> Only if ?\RET, ?\TAB, etc. work too (otherwise it's just too
> inconsistent -- if ?\SPC worked then people would quite naturally expect
> things like ?\RET or ?\TAB to work as well).

I agree, so I redraw my proposal!

Thinking about it, ?\s is a good choice after all!

As Luc points out, we can introduce the \s for space without any risk
of breaking existing code, since the only valid syntax for using \s is
when it is followed by a -, and \s-... doesn't even work in strings.

So I suggest that ?\s is a space unless the `s' is immediately followed
by `-' (e.g. ?\s-a) in which case it has the usual meaning as a super
modifier.

In strings, "\s" should also mean a space unless it is following by
a `-' (as in "\s-a") which should produce an error as it does now.

To follow a space with a `-' in a string use " -" as before, or "\s\-".

The following patch does this:

Index: lread.c
===================================================================
RCS file: /cvsroot/emacs/emacs/src/lread.c,v
retrieving revision 1.305
diff -c -r1.305 lread.c
*** lread.c	4 Feb 2003 14:03:12 -0000	1.305
--- lread.c	6 Feb 2003 10:37:54 -0000
***************
*** 1698,1705 ****
  
      case 's':
        c = READCHAR;
!       if (c != '-')
! 	error ("Invalid escape character syntax");
        c = READCHAR;
        if (c == '\\')
  	c = read_escape (readcharfun, 0, byterep);
--- 1698,1707 ----
  
      case 's':
        c = READCHAR;
!       if (c != '-') {
! 	UNREAD (c);
! 	return ' ';
!       }
        c = READCHAR;
        if (c == '\\')
  	c = read_escape (readcharfun, 0, byterep);


-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: ?\_ patch

[Luc Teirlinck]

Richard and Handa seem to object against any change in Emacs for the
purpose.  I do not know whether or not that is due to the fact that
they might believe the required change to be more substantial than it
actually is (as can be seen from your patch).  Of course, in addition
to your patch some "secondary" changes to support the primary change
would be necessary, as can be seen from Ted's original patch.
However, such secondary (mostly documentation) changes are necessary
anyway if we are going to alert people to the dangers of ?\ in terms
of trailing whitespace removal and tabification, and propose
alternatives.

If you compare ?\s with ?\040 or ?\x20 (which I consider so closely
related that I consider them as one solution), then:

?\s:
Disadvantages:
1. Possible confusion with ?\s-
2. Requires change in Emacs.
Advantages:
1. Intuitive and concise.
2. Required change in Emacs is limited.

?\040 (or ?\x20)
Disadvantages:
1. Relies on ASCII ordering.
2. Not terribly intuitive.
Advantage:
No change in Emacs required.

I believe that the third suggested solution, (string-to-car " "),
presents some problems if one systematically would want to replace \?
everywhere with it, as I pointed out before.

The main objection against ?\s seems to be the change in Emacs.
In as far as ?\040 (or ?\x20) are concerned, Handa does not seem to
object against the reliance on ASCII and he knows that stuff a lot
better than I do.

So it all comes down to whether the gain in intuitiveness from \?s is
worth the (small) change in Emacs.  Of course, intuitiveness is a
relative concept.

Sincerely,

Luc.

Re: ?\_ patch

[Miles Bader]

Luc Teirlinck <teirllm@dms.auburn.edu> writes:
> ?\s:
> Disadvantages:
> 1. Possible confusion with ?\s-
> 2. Requires change in Emacs.
> Advantages:
> 1. Intuitive and concise.
> 2. Required change in Emacs is limited.
> 
> ?\040 (or ?\x20)
> Disadvantages:
> 1. Relies on ASCII ordering.
> 2. Not terribly intuitive.
> Advantage:
> No change in Emacs required.

?\_:
Disadvantages:
1. Requires change in Emacs.
Advantages:
1. Intuitive and concise.
2. Required change in Emacs is limited.

I.e., the same as ?\s except with fewer disadvantages...

-Miles
-- 
80% of success is just showing up.  --Woody Allen

Re: ?\_ patch

[Luc Teirlinck]

Miles Bader wrote:

   ?\_:
   Disadvantages:
   1. Requires change in Emacs.
   Advantages:
   1. Intuitive and concise.
   2. Required change in Emacs is limited.

   I.e., the same as ?\s except with fewer disadvantages...

As Handa pointed out, ?\_ would have broken tons of existing code,
which, for whatever reason, used ?\_ instead of ?_  Breaking existing
code is a big disadvantage.

Sincerely,

Luc.

Re: ?\_ patch

[Miles Bader]

Luc Teirlinck <teirllm@dms.auburn.edu> writes:
> As Handa pointed out, ?\_ would have broken tons of existing code,
> which, for whatever reason, used ?\_ instead of ?_  Breaking existing
> code is a big disadvantage.

I didn't see Handa's message (there was a wierd burp in the emacs-devel
mailing list, so perhaps it got nuked there...), but I did a grep and
you're right -- a bunch of code uses ?\_ to mean ?_ .

I can't imagine _why_ people did that, but they did, so I guess ?\_ is
no good.  Bleah.

I really don't like the wierd double identity of ?\s, but I'm not sure
what alternative there is... Perhaps kim's idea (?\SPC), but with a less
ambiguous syntax, e.g., #\SPC or something.

Does xemacs do anything about this issue?

-Miles
-- 
Fast, small, soon; pick any 2.

Re: ?\_ patch

[Kenichi Handa]

In article <buo65rwbibj.fsf@mcspd15.ucom.lsi.nec.co.jp>, Miles Bader <miles@lsi.nec.co.jp> writes:

> I really don't like the wierd double identity of ?\s, but I'm not sure
> what alternative there is... Perhaps kim's idea (?\SPC), but with a less
> ambiguous syntax, e.g., #\SPC or something.

Ah!  I completely forgot about `#' syntax.  I think it's a
good idea.

Currently single-key-description produces these: TAB, RET,
ESC, SPC, DEL.  Allowing the corresponding #\TAB, #\RET,
#\ESC, #\SPC, #\DEL seems nifty.

---
Ken'ichi HANDA
handa@m17n.org

Re: ?\_ patch

[Richard Stallman]

    So I suggest that ?\s is a space unless the `s' is immediately followed
    by `-' (e.g. ?\s-a) in which case it has the usual meaning as a super
    modifier.

    In strings, "\s" should also mean a space unless it is following by
    a `-' (as in "\s-a") which should produce an error as it does now.

I have a bad feeling about this complication in strings, because it would
mean introducing a feature that doesn't always work the way people
would expect it to work.

Perhaps we could introduce the \s feature only for character constants.
In that particular context, there is no ambiguity.

Re: ?\_ patch

[Kim F. Storm]

Richard Stallman <rms@gnu.org> writes:

>     So I suggest that ?\s is a space unless the `s' is immediately followed
>     by `-' (e.g. ?\s-a) in which case it has the usual meaning as a super
>     modifier.
> 
>     In strings, "\s" should also mean a space unless it is following by
>     a `-' (as in "\s-a") which should produce an error as it does now.
> 
> I have a bad feeling about this complication in strings, because it would
> mean introducing a feature that doesn't always work the way people
> would expect it to work.

I agree.

Here's what the Elisp manual says:

   Strings cannot hold characters that have the hyper, super or alt
   modifiers; ...

Since there is definitely NO code today which relies on being able to
specify a super modifier like "\s-a" in a string (emacs will barf with
an "Invalid modifier in string" error if you try), we could simply
define \s to ALWAYS give a space in a string.

Then \s will always work the way people expect!

> 
> Perhaps we could introduce the \s feature only for character constants.
> In that particular context, there is no ambiguity.

If we remove the (non-functional) special meaning of \s in strings as
suggested above, there's no ambiguity for strings either.

Here is a modified patch:

Index: lread.c
===================================================================
RCS file: /cvsroot/emacs/emacs/src/lread.c,v
retrieving revision 1.305
diff -c -r1.305 lread.c
*** lread.c	4 Feb 2003 14:03:12 -0000	1.305
--- lread.c	7 Feb 2003 09:39:46 -0000
***************
*** 1697,1705 ****
        return c | alt_modifier;
  
      case 's':
        c = READCHAR;
!       if (c != '-')
! 	error ("Invalid escape character syntax");
        c = READCHAR;
        if (c == '\\')
  	c = read_escape (readcharfun, 0, byterep);
--- 1697,1709 ----
        return c | alt_modifier;
  
      case 's':
+       if (stringp)
+ 	return ' ';
        c = READCHAR;
!       if (c != '-') {
! 	UNREAD (c);
! 	return ' ';
!       }
        c = READCHAR;
        if (c == '\\')
  	c = read_escape (readcharfun, 0, byterep);




BTW, why is `backslash space' ignored in a string?
        "\ " => ""

-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: ?\_ patch

[Andreas Schwab]

storm@cua.dk (Kim F. Storm) writes:

|> BTW, why is `backslash space' ignored in a string?
|>         "\ " => ""

For writing things like "\xdead\ beef" .

Andreas.

-- 
Andreas Schwab, SuSE Labs, schwab@suse.de
SuSE Linux AG, Deutschherrnstr. 15-19, D-90429 Nürnberg
Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: ?\_ patch

[Miles Bader]

On Fri, Feb 07, 2003 at 03:17:33PM +0100, Kim F. Storm wrote:
> > > I really don't like the wierd double identity of ?\s, but I'm not sure
> > > what alternative there is... Perhaps kim's idea (?\SPC), but with a less
> > > ambiguous syntax, e.g., #\SPC or something.
> > 
> > Ah!  I completely forgot about `#' syntax.  I think it's a
> > good idea.
> 
> I don't object to it, but I think it is unnecessary.
> 
> Why add a whole new syntax just because we need ?\s which can be added
> with no hazzle at all?

Because #\SPC is actually pretty self-explanatory whereas ?\s isn't.

> If everyone else agrees this is a good idea, I'd suggest
> using #?SPC, #?TAB, etc.

I thought about that too, but I think #\SPC is better, because the `\' leaves
a bit of whitespace between itself and the following character so the `SPC'
stands out quite distinctly.  `?' on the other isn't visually distinct, so
#?SPC looks like a bit of a muddle.

-Miles
-- 
"1971 pickup truck; will trade for guns"

Re: ?\_ patch

[Miles Bader]

On Fri, Feb 07, 2003 at 03:52:02PM +0100, Kim F. Storm wrote:
> > Because #\SPC is actually pretty self-explanatory whereas ?\s isn't.
> 
> IMO, \s is just as self-explanatory as \t, \n, and \r.

The latter are `self-explanatory' only because they're used in C, and so are
very familar to programmers.  \s is not.

> > I thought about that too, but I think #\SPC is better, because the `\'
> > leaves a bit of whitespace between itself and the following character so
> > the `SPC' stands out quite distinctly.  `?' on the other isn't visually
> > distinct, so #?SPC looks like a bit of a muddle.
> 
> Then what about simply using ?SPC, ?TAB, etc.

No, you're missing the point.  My object to `?' is that the `?' is not
visually distinct from the `S' -- they tend to `run into' each other. 
`\', on the other hand has a bunch of whitespace on the right side of it's
glyph, and so is much more visually distinct from the following character.

So getting rid of the `#' doesn't help at all.

Morever, `#' is _good_ becaues it's the general lisp syntax for special
syntax, so a lisp programmer will be much more likely to realize what's going
on (if he's unfamiliar with this particular bit of syntax) if he sees #\SPC
than if he sees ?SPC.

> Can anyone think of existing code which will be broken by that
> approach?

I don't know, but I think it doesn't matter; we should just use a `#' escape.

-Miles
-- 
I'd rather be consing.

Re: ?\_ patch

[Kim F. Storm]

Kenichi Handa <handa@m17n.org> writes:

> In article <buo65rwbibj.fsf@mcspd15.ucom.lsi.nec.co.jp>, Miles Bader <miles@lsi.nec.co.jp> writes:
> 
> > I really don't like the wierd double identity of ?\s, but I'm not sure
> > what alternative there is... Perhaps kim's idea (?\SPC), but with a less
> > ambiguous syntax, e.g., #\SPC or something.
> 
> Ah!  I completely forgot about `#' syntax.  I think it's a
> good idea.

I don't object to it, but I think it is unnecessary.

Why add a whole new syntax just because we need ?\s which can be added
with no hazzle at all?

> 
> Currently single-key-description produces these: TAB, RET,
> ESC, SPC, DEL.  Allowing the corresponding #\TAB, #\RET,
> #\ESC, #\SPC, #\DEL seems nifty.

If everyone else agrees this is a good idea, I'd suggest
using #?SPC, #?TAB, etc.

-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: ?\_ patch

[Dmitry Paduchikh]

    Hi!

>>>>> Miles Bader <miles@gnu.org> writes:

> Morever, `#' is _good_ becaues it's the general lisp syntax for special
> syntax, so a lisp programmer will be much more likely to realize what's going
> on (if he's unfamiliar with this particular bit of syntax) if he sees #\SPC
> than if he sees ?SPC.

If you will do this, why not introduce such alternative syntax for
_all_ characters? It would give the advantage of being consistent with
other lispish languages since both CL and Scheme use `#\' to denote
characters.

Re: ?\_ patch

[Kim F. Storm]

Miles Bader <miles@gnu.org> writes:

> On Fri, Feb 07, 2003 at 03:17:33PM +0100, Kim F. Storm wrote:
> > > > I really don't like the wierd double identity of ?\s, but I'm not sure
> > > > what alternative there is... Perhaps kim's idea (?\SPC), but with a less
> > > > ambiguous syntax, e.g., #\SPC or something.
> > > 
> > > Ah!  I completely forgot about `#' syntax.  I think it's a
> > > good idea.
> > 
> > I don't object to it, but I think it is unnecessary.
> > 
> > Why add a whole new syntax just because we need ?\s which can be added
> > with no hazzle at all?
> 
> Because #\SPC is actually pretty self-explanatory whereas ?\s isn't.

IMO, \s is just as self-explanatory as \t, \n, and \r.

> 
> > If everyone else agrees this is a good idea, I'd suggest
> > using #?SPC, #?TAB, etc.
> 
> I thought about that too, but I think #\SPC is better, because the `\' leaves
> a bit of whitespace between itself and the following character so the `SPC'
> stands out quite distinctly.  `?' on the other isn't visually distinct, so
> #?SPC looks like a bit of a muddle.
> 

Then what about simply using ?SPC, ?TAB, etc.

Today, if I eval (using C-j) the following in *scratch*, I get:

        ?S       => 83
        ?SPC     => 83
        '(?SPC)  => (83 PC)

but if I do
        M-: ? S P C RET
I get
        list: Trailing garbage following expression

so the current character syntax really is pretty obscure.

This could easily be changed to:

        ?S       => 83
        ?SPC     => 32
        '(?SPC)  => (32)

Can anyone think of existing code which will be broken by that
approach?

-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: ?\_ patch

[Damien Elmes]

storm@cua.dk (Kim F. Storm) writes:

> The Lisp manual doesn't refer to C syntax to explain e.g. ?\t.
> So to me, \s and \t are equally self-explanatory.

I agree. I also think that the more verbose you make the "proper" way to go
about it, the more likely people are going to be lazy and just use "? ". Why
introduce a whole new #\ syntax when ?foo is currently widely used to
represent characters?

Regards,
-- 
Damien Elmes

Re: ?\_ patch

[Kim F. Storm]

Miles Bader <miles@gnu.org> writes:

> On Fri, Feb 07, 2003 at 03:52:02PM +0100, Kim F. Storm wrote:
> > > Because #\SPC is actually pretty self-explanatory whereas ?\s isn't.
> > 
> > IMO, \s is just as self-explanatory as \t, \n, and \r.
> 
> The latter are `self-explanatory' only because they're used in C, and so are
> very familar to programmers.  \s is not.

The Lisp manual doesn't refer to C syntax to explain e.g. ?\t.
So to me, \s and \t are equally self-explanatory.

On the other hand, I don't find #\SPC self-explanatory (but ?\SPC is).

> 
> > > I thought about that too, but I think #\SPC is better, because the `\'
> > > leaves a bit of whitespace between itself and the following character so
> > > the `SPC' stands out quite distinctly.  `?' on the other isn't visually
> > > distinct, so #?SPC looks like a bit of a muddle.
> > 
> > Then what about simply using ?SPC, ?TAB, etc.
> 
> No, you're missing the point.  My object to `?' is that the `?' is not
> visually distinct from the `S' -- they tend to `run into' each other. 
> `\', on the other hand has a bunch of whitespace on the right side of it's
> glyph, and so is much more visually distinct from the following character.

Ok, I agree.  However, I still think it is best to keep the ? as part
of the read syntax for a character, i.e. I prefer ?\SPC to #\SPC as
the ?\ prefix clearly identifies this as a character, e.g. I find the
use of ? in the following example more consistent than #:

        (memq ch '(?\SPC ?\- ?\n ?\t))
        (memq ch '(#\SPC ?\- ?\n ?\t))

I just grepped for "[?][\\][A-Z][^-]" in all of lisp/ and there isn't
a single occurrence.  So the ?\XXX syntax should be safe.

> Morever, `#' is _good_ becaues it's the general lisp syntax for special
> syntax, so a lisp programmer will be much more likely to realize what's going
> on (if he's unfamiliar with this particular bit of syntax) if he sees #\SPC
> than if he sees ?SPC.

But `?' is _much better_ than `#' here, since _any_ lisp programmer
should know that this denotes a character, whereas `#' just denotes it
as "special" (but there's nothing special about a space character?).

I vote for ?\SPC over #\SPC.

But I still think adding ?\s is the best choice; it's simple, it
doesn't break anything, and it doesn't require us to introduce a
completely new character syntax.

-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: ?\_ patch

[John Paul Wallington]

Kim F. Storm wrote:

>  What is the CL and Scheme syntax for a space character ?

#\Space

Re: ?\_ patch

[Kim F. Storm]

Dmitry Paduchikh <paduch@imm.uran.ru> writes:

>     Hi!
> 
> >>>>> Miles Bader <miles@gnu.org> writes:
> 
> > Morever, `#' is _good_ becaues it's the general lisp syntax for special
> > syntax, so a lisp programmer will be much more likely to realize what's going
> > on (if he's unfamiliar with this particular bit of syntax) if he sees #\SPC
> > than if he sees ?SPC.
> 
> If you will do this, why not introduce such alternative syntax for
> _all_ characters? It would give the advantage of being consistent with
> other lispish languages since both CL and Scheme use `#\' to denote
> characters.

What is the CL and Scheme syntax for a space character ?

-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: ?\_ patch

[Dmitry Paduchikh]

>>>>> Kim F Storm <no-spam@cua.dk> writes:

> What is the CL and Scheme syntax for a space character ?

Standard syntax for space in CL is `#\ ' or `#\Space'. See
<http://www.lispworks.com/reference/HyperSpec/Body/02_dha.htm> for
details.

Re: ?\_ patch

[John Paul Wallington]

Kim F. Storm wrote:

> > >  What is the CL and Scheme syntax for a space character ?
> > 
> > #\Space
> > 
>  
>  So I guess introducing #\SPC for space in emacs would be even more
>  confusing than ?\s or ?\SPC.

CL, Scheme, and other dialects like rep support #\ syntax for chars
where you can have symbolic names like space or newline.

Maybe emacs lisp could do likewise.

Re: ?\_ patch

[Kim F. Storm]

John Paul Wallington <jpw@gnu.org> writes:

> Kim F. Storm wrote:
> 
> >  What is the CL and Scheme syntax for a space character ?
> 
> #\Space
> 

So I guess introducing #\SPC for space in emacs would be even more
confusing than ?\s or ?\SPC.

-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: ?\_ patch

[Luc Teirlinck]

Damien Elmes wrote:  
   
   storm@cua.dk (Kim F. Storm) writes:

   > The Lisp manual doesn't refer to C syntax to explain e.g. ?\t.
   > So to me, \s and \t are equally self-explanatory.

   I agree. I also think that the more verbose you make the "proper"
   way to go about it, the more likely people are going to be lazy and
   just use "? ". Why introduce a whole new #\ syntax when ?foo is
   currently widely used to represent characters?

I agree with Damien and Kim on this.  I believe that the few times
that one actually needs ?\ are not sufficient to introduce an entirely
new style of character syntax into emacs.  This seems like the
proverbial sledgehammer to crush an almond.  I do not believe that we
should try to emulate Common Lisp in this particular respect.  I like
Common Lisp, but its treatment of characters is quite simply not its
strongest point.  Whenever I have to play around with characters, I
would much rather use Elisp.

I prefer to go for Kim's latest ?\s version (where \s denotes a space
even in "\s-a", since the super modifier is invalid inside strings
anyway).  Short of that, I would rather stay with ?\040 or ?\x20 than
to introduce entirely new-style character syntax.

Sincerely,

Luc.

Re: ?\_ patch

[Miles Bader]

On Fri, Feb 07, 2003 at 05:50:21PM +0100, Kim F. Storm wrote:
> > >  What is the CL and Scheme syntax for a space character ?
> > 
> > #\Space
> 
> So I guess introducing #\SPC for space in emacs would be even more
> confusing than ?\s or ?\SPC.

Oh, don't be silly.  _all_ of the above are pretty clear to anyone with even
a modicum of clue, except ?\s.

-miles
-- 
`Cars give people wonderful freedom and increase their opportunities.
 But they also destroy the environment, to an extent so drastic that
 they kill all social life' (from _A Pattern Language_)

Re: ?\_ patch

[Kim F. Storm]

Miles Bader <miles@gnu.org> writes:

> On Fri, Feb 07, 2003 at 05:50:21PM +0100, Kim F. Storm wrote:
> > > >  What is the CL and Scheme syntax for a space character ?
> > > 
> > > #\Space
> > 
> > So I guess introducing #\SPC for space in emacs would be even more
> > confusing than ?\s or ?\SPC.
> 
> Oh, don't be silly.  _all_ of the above are pretty clear to anyone with even
> a modicum of clue, except ?\s.

IMO, it would be silly to adapt the CL and Scheme syntax for
characters, but use different names for the characters.  If we want to
use #\<something> to denote a space, it should be #\Space - not #\SPC.


-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: ?\_ patch

[Richard Stallman]

    Morever, `#' is _good_ becaues it's the general lisp syntax for special
    syntax, so a lisp programmer will be much more likely to realize what's going
    on (if he's unfamiliar with this particular bit of syntax) if he sees #\SPC
    than if he sees ?SPC.

? is already our syntax for character constants; I see no reason
for changing that.

Re: ?\_ patch

[Miles Bader]

storm@cua.dk (Kim F. Storm) writes:
> IMO, it would be silly to adapt the CL and Scheme syntax for
> characters, but use different names for the characters.  If we want to
> use #\<something> to denote a space, it should be #\Space - not #\SPC.

It should be both (no reason not to)...

-Miles
-- 
The secret to creativity is knowing how to hide your sources.
  --Albert Einstein

Re: ?\_ patch

[Richard Stallman]

	  case 's':
    +       if (stringp)
    + 	return ' ';
	    c = READCHAR;
    !       if (c != '-') {
    ! 	UNREAD (c);
    ! 	return ' ';
    !       }
	    c = READCHAR;
	    if (c == '\\')

This might be a good idea if we combine it with the feature that it is
an error to put a digit or a symbol character or # after a character
constant with no space in between.

Re: ?\_ patch

[Andreas Schwab]

storm@cua.dk (Kim F. Storm) writes:

|> IMO, \s is just as self-explanatory as \t, \n, and \r.

IMHO, \s is bad because it has already two other meanings, one as the
super prefix, another as the syntax matcher in a regex.  Both have nothing
to do with spaces.

Andreas.

-- 
Andreas Schwab, SuSE Labs, schwab@suse.de
SuSE Linux AG, Deutschherrnstr. 15-19, D-90429 Nürnberg
Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: ?\_ patch

[Kim F. Storm]

Richard Stallman <rms@gnu.org> writes:

> 	  case 's':
>     +       if (stringp)
>     + 	return ' ';
> 	    c = READCHAR;
>     !       if (c != '-') {
>     ! 	UNREAD (c);
>     ! 	return ' ';
>     !       }
> 	    c = READCHAR;
> 	    if (c == '\\')
> 
> This might be a good idea if we combine it with the feature that it is
> an error to put a digit or a symbol character or # after a character
> constant with no space in between.

There are several places where something like '(? ?...) or '(?\ ?...)
is used, i.e. where no additional space separates the "? " character
from the "?" of the next character.  

So we would at least need to make an exception that no additional
space is needed after a (real) space.

-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: ?\_ patch

[Richard Stallman]

    > This might be a good idea if we combine it with the feature that it is
    > an error to put a digit or a symbol character or # after a character
    > constant with no space in between.

    There are several places where something like '(? ?...) or '(?\ ?...)
    is used, i.e. where no additional space separates the "? " character
    from the "?" of the next character.  

Looking closely, there is no overlap between these cases and the ones
I said should be an error.

Re: ?\_ patch

[Kim F. Storm]

Richard Stallman <rms@gnu.org> writes:

>     > This might be a good idea if we combine it with the feature that it is
>     > an error to put a digit or a symbol character or # after a character
>     > constant with no space in between.
> 
>     There are several places where something like '(? ?...) or '(?\ ?...)
>     is used, i.e. where no additional space separates the "? " character
>     from the "?" of the next character.  
> 
> Looking closely, there is no overlap between these cases and the ones
> I said should be an error.

Ok.  So I assume that you want ?XY to be invalid syntax
unless Y is <= 040 or one of  " ' ; ( ) [ ] ? or #

-- 
Kim F. Storm  http://www.cua.dk

Re: ?\_ patch

[Richard Stallman]

    Ok.  So I assume that you want ?XY to be invalid syntax
    unless Y is <= 040 or one of  " ' ; ( ) [ ] ? or #

That is the idea.  I can see how someone could have written (?a(b))
and think it legitimate, but a person could hardly argue that (?ab)
instead of (?a b) is legitimate.

It's possible there is some other syntax char I have forgotten
that should be legitimate there too.  If you can think of one,
please add it.

Re: ?\_ patch

[Kim F. Storm]

Richard Stallman <rms@gnu.org> writes:

>     Ok.  So I assume that you want ?XY to be invalid syntax
>     unless Y is <= 040 or one of  " ' ; ( ) [ ] ? or #
> 
> That is the idea.  I can see how someone could have written (?a(b))
> and think it legitimate, but a person could hardly argue that (?ab)
> instead of (?a b) is legitimate.
> 
> It's possible there is some other syntax char I have forgotten
> that should be legitimate there too.  If you can think of one,
> please add it.

I have installed the changes.

Edward O'Connor suggested to modify checkdoc to look for ?\ at the end
of lines and replace them with the new syntax.  That's a good idea,
but I have not done this.

-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk