[cjm@pobox.com: Case-insensitive partial-completion bug]

[cjm@pobox.com: Case-insensitive partial-completion bug]

[Richard Stallman]

Would someone please DTRT and ack?

------- Start of forwarded message -------
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
From: "Christopher J. Madsen" <cjm@pobox.com>
To: emacs-pretest-bug@gnu.org
Date: Thu, 09 Nov 2006 10:47:45 -0600 (CST)
Subject: Case-insensitive partial-completion bug
X-Spam-Status: No, score=0.0 required=5.0 tests=none autolearn=failed 
	version=3.0.4

Please describe exactly what actions triggered the bug
and the precise symptoms of the bug:

In a shell, create a new directory and cd there.
> touch Foo Foobar
> emacs -Q

In the *scratch* buffer, type:
(setq read-file-name-completion-ignore-case t) C-j

M-x partial-completion-mode <return>

C-x C-f foo <tab>


I expected it to change "foo" to "Foo", because all possible
completions begin with "Foo".  That's what it does if you don't enable
partial-completion-mode.  But with partial-completion-mode enabled, it
stays "foo" and Emacs displays the list of possible completions.

If you type "b <tab>", it will change "foob" to "Foobar" (as expected).

This used to work in Emacs 21.3.1 (although there the variable is
completion-ignore-case instead).  I first noticed the bug a couple
months ago, but it's still present in the CVS HEAD version I compiled
yesterday.


In GNU Emacs 22.0.90.1 (i686-pc-linux-gnu, GTK+ Version 2.8.19)
 of 2006-11-08 on bit
X server distributor `The X.Org Foundation', version 11.0.70101000
configured using `configure '--prefix=/usr' '--host=i686-pc-linux-gnu' '--mandir=/usr/share/man' '--infodir=/usr/share/info' '--datadir=/usr/share' '--sysconfdir=/etc' '--localstatedir=/var/lib' '--program-suffix=.emacs-22.0.90' '--without-carbon' '--with-x' '--with-xpm' '--with-toolkit-scroll-bars' '--with-jpeg' '--with-tiff' '--with-gif' '--with-png' '--with-x-toolkit=gtk' '--build=i686-pc-linux-gnu' 'CFLAGS=-O2' 'build_alias=i686-pc-linux-gnu' 'host_alias=i686-pc-linux-gnu''

Important settings:
  value of $LC_ALL: en_US
  value of $LC_COLLATE: nil
  value of $LC_CTYPE: nil
  value of $LC_MESSAGES: nil
  value of $LC_MONETARY: nil
  value of $LC_NUMERIC: nil
  value of $LC_TIME: nil
  value of $LANG: nil
  locale-coding-system: iso-8859-1
  default-enable-multibyte-characters: t

Major mode: Lisp Interaction

Minor modes in effect:
  partial-completion-mode: t
  tooltip-mode: t
  tool-bar-mode: t
  mouse-wheel-mode: t
  menu-bar-mode: t
  file-name-shadow-mode: t
  global-font-lock-mode: t
  font-lock-mode: t
  blink-cursor-mode: t
  unify-8859-on-encoding-mode: t
  utf-translate-cjk-mode: t
  auto-compression-mode: t
  line-number-mode: t

Recent input:
( s e t q SPC r e a d - f i l e <escape> <tab> - c
<escape> <tab> SPC t ) C-j M-x p a r t <tab> <return>
C-x C-f f o o <tab> C-g M-x r e p o r t <tab> <return>

Recent messages:
(emacs -Q)
For information about the GNU Project and its goals, type C-h C-p.
Loading complete...
Loading advice...done
Loading complete...done
Partial-Completion mode enabled
Quit
Loading emacsbug...
Loading regexp-opt...done
Loading emacsbug...done



_______________________________________________
emacs-pretest-bug mailing list
emacs-pretest-bug@gnu.org
http://lists.gnu.org/mailman/listinfo/emacs-pretest-bug
------- End of forwarded message -------

Re: [cjm@pobox.com: Case-insensitive partial-completion bug]

[Markus Triska]

Richard Stallman <rms@gnu.org> writes:

> Would someone please DTRT and ack?

Intentional feature:

revision 1.54
date: 2006-06-23 15:05:03 +0200;  author: eliz;  state: Exp;  lines: +50 -20
(PC-do-completion): Retain capitalization of user input, when possible, even
if completion-ignore-case is set.

Re: [cjm@pobox.com: Case-insensitive partial-completion bug]

[Richard Stallman]

The complaint was:

    I expected it to change "foo" to "Foo", because all possible
    completions begin with "Foo".  That's what it does if you don't enable
    partial-completion-mode.  But with partial-completion-mode enabled, it
    stays "foo" and Emacs displays the list of possible completions.

Michael Ernst wrote the patch which implemented this behavior.

Michael, why is it desirable for partial-completion mode to handle
this case differently from ordinary completion?

Re: [cjm@pobox.com: Case-insensitive partial-completion bug]

[Michael Ernst]

> The complaint was:
> 
>     I expected it to change "foo" to "Foo", because all possible
>     completions begin with "Foo".  That's what it does if you don't enable
>     partial-completion-mode.  But with partial-completion-mode enabled, it
>     stays "foo" and Emacs displays the list of possible completions.
> 
> Michael Ernst wrote the patch which implemented this behavior.
> 
> Michael, why is it desirable for partial-completion mode to handle
> this case differently from ordinary completion?

Thanks for forwarding the bug report.  I'm aware of this problem, and I
agree it should be fixed, but I haven't yet found time to correct it
(though it remains on my to-do list).  I have noticed it once in the 6
months I have been using the new code, and the new code corrects problems
that I noticed much more frequently.

My notes state that the problem is near the "same char (modulo case); no
action" comment, and near

		  ;; Retain capitalization of user input even if
		  ;; completion-ignore-case is set.

If the list `poss' has many possibilities with different capitalization,
then we want to retain the capitalization as input by the user.  However,
if they are all the same, we want to adjust the capitalization.

Below my signature is a test case that illustrates the problem in a
reproducible way.

                    -Michael Ernst
                     mernst@alum.mit.edu



(defun completion-test-4 (initial-input &optional ignore-case)
  (let ((completion-ignore-case ignore-case))
    (completing-read "Input: "
                     '(("getMillis" . 1)
                       ("getMillisToDecideToPopup" . 2)
                       ("getMillisToPopup" . 3)
                       ("getMillisecond" . 4))
                     nil                ; no predicate, all answers acceptable
                     t                  ; require match
                     initial-input
                     )))

(partial-completion-mode 1)
(setq PC-meta-flag t)

(completion-test-4 "getmilli" t)
;; user types:  M-TAB    [normal completion]
;; result:  getMillis    [chooses the element that matches modulo case]
;; user types:  RET      [partial completion]
;; result:  getmillis    [problem:  all lower-case, second RET doesn't select]

;; Actually, it seems like a problem that normal completion chooses the
;; matching element immediately rather than giving a list of all
;; possibilities.  Is this a bug in normal completion?

Re: [cjm@pobox.com: Case-insensitive partial-completion bug]

[Richard Stallman]

We are going to release Emacs 22 soon.  Would you please make time to
fix this bug?

I hope you'll respond to this message with a fix.

Re: [cjm@pobox.com: Case-insensitive partial-completion bug]

[Michael Ernst]

> We are going to release Emacs 22 soon.  Would you please make time to
> fix this bug?

I will try to find the time.  (The code is undocumented and poorly
structured, so it is tricky to make changes without negative effects.)

I'm glad that Emacs 22 will be released soon.

                    -Michael Ernst
                     mernst@alum.mit.edu

Re: [cjm@pobox.com: Case-insensitive partial-completion bug]

[Richard Stallman]

    I will try to find the time.  (The code is undocumented and poorly
    structured, so it is tricky to make changes without negative effects.)

How about first adding some comments?

Re: [cjm@pobox.com: Case-insensitive partial-completion bug]

[Johan Bockgård]

I hope the bug below will get fixed. I run into it quite frequently
when reading Info:

  emacs -Q -f partial-completion-mode -f info-emacs-manual

  `g mark RET' doesn't select and exit minibuffer.


On Mon, Nov 20 2006, Michael Ernst wrote:

>> The complaint was:
>> 
>>     I expected it to change "foo" to "Foo", because all possible
>>     completions begin with "Foo".  That's what it does if you don't enable
>>     partial-completion-mode.  But with partial-completion-mode enabled, it
>>     stays "foo" and Emacs displays the list of possible completions.
>> 
>> Michael Ernst wrote the patch which implemented this behavior.
>> 
>> Michael, why is it desirable for partial-completion mode to handle
>> this case differently from ordinary completion?
>
> Thanks for forwarding the bug report.  I'm aware of this problem, and I
> agree it should be fixed, but I haven't yet found time to correct it
> (though it remains on my to-do list).  I have noticed it once in the 6
> months I have been using the new code, and the new code corrects problems
> that I noticed much more frequently.
>
> My notes state that the problem is near the "same char (modulo case); no
> action" comment, and near
>
> 		  ;; Retain capitalization of user input even if
> 		  ;; completion-ignore-case is set.
>
> If the list `poss' has many possibilities with different capitalization,
> then we want to retain the capitalization as input by the user.  However,
> if they are all the same, we want to adjust the capitalization.
>
> Below my signature is a test case that illustrates the problem in a
> reproducible way.
>
>                     -Michael Ernst
>                      mernst@alum.mit.edu
>
>
>
> (defun completion-test-4 (initial-input &optional ignore-case)
>   (let ((completion-ignore-case ignore-case))
>     (completing-read "Input: "
>                      '(("getMillis" . 1)
>                        ("getMillisToDecideToPopup" . 2)
>                        ("getMillisToPopup" . 3)
>                        ("getMillisecond" . 4))
>                      nil                ; no predicate, all answers acceptable
>                      t                  ; require match
>                      initial-input
>                      )))
>
> (partial-completion-mode 1)
> (setq PC-meta-flag t)
>
> (completion-test-4 "getmilli" t)
> ;; user types:  M-TAB    [normal completion]
> ;; result:  getMillis    [chooses the element that matches modulo case]
> ;; user types:  RET      [partial completion]
> ;; result:  getmillis    [problem:  all lower-case, second RET doesn't select]
>
> ;; Actually, it seems like a problem that normal completion chooses the
> ;; matching element immediately rather than giving a list of all
> ;; possibilities.  Is this a bug in normal completion?

Re: [cjm@pobox.com: Case-insensitive partial-completion bug]

[Richard Stallman]

    I hope the bug below will get fixed. I run into it quite frequently
    when reading Info:

      emacs -Q -f partial-completion-mode -f info-emacs-manual

      `g mark RET' doesn't select and exit minibuffer.

I don't use Partial Completion mode.  Is this a bug in it?
It looks that way, but I can't be sure.

If it is, does anyone want to fix it?

Re: [cjm@pobox.com: Case-insensitive partial-completion bug]

[Chong Yidong]

Richard Stallman <rms@gnu.org> writes:

>     I hope the bug below will get fixed. I run into it quite frequently
>     when reading Info:
>
>       emacs -Q -f partial-completion-mode -f info-emacs-manual
>
>       `g mark RET' doesn't select and exit minibuffer.
>
> I don't use Partial Completion mode.  Is this a bug in it?
> It looks that way, but I can't be sure.
>
> If it is, does anyone want to fix it?

I think this is an oversight in the logic of `PC-do-completion'.  The
following patch should fix it.

*** emacs/lisp/complete.el.~1.59.~	2007-01-21 08:38:26.000000000 -0500
--- emacs/lisp/complete.el	2007-03-04 13:28:25.000000000 -0500
***************
*** 404,411 ****
  
      ;; Check if buffer contents can already be considered complete
      (if (and (eq mode 'exit)
! 	     (test-completion-ignore-case str table pred))
! 	'complete
  
        ;; Do substitutions in directory names
        (and filename
--- 404,418 ----
  
      ;; Check if buffer contents can already be considered complete
      (if (and (eq mode 'exit)
! 	     (test-completion str table pred))
! 	(progn
! 	  ;; If completion-ignore-case is non-nil, insert the
! 	  ;; completion string since that may have a different case.
! 	  (when completion-ignore-case
! 	    (setq str (try-completion str table pred))
! 	    (delete-region beg end)
! 	    (insert str))
! 	  'complete)
  
        ;; Do substitutions in directory names
        (and filename