flyspell interferes with C-c C-c (mail mode, vc-log-mode, etc)

flyspell interferes with C-c C-c (mail mode, vc-log-mode, etc)

[era eriksson]

This bug report will be sent to the Free Software Foundation,
 not to your local site managers!!
Please write in English, because the Emacs maintainers do not have
translators to read other languages for them.

In GNU Emacs 20.7.2 (i386-debian-linux-gnu, X toolkit)
 of Fri Apr  5 2002 on cyberhq modified by Debian
configured using `configure  i386-debian-linux-gnu --prefix=/usr --sharedstatedir=/var/lib --libexecdir=/usr/lib --localstatedir=/var/lib --infodir=/usr/share/info --mandir=/usr/share/man --with-pop=yes --with-x=yes --with-x-toolkit=yes'

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

It seems that the keystroke C-c C-c is intercepted and discarded when
Flyspell is active and the token under the text cursor is misspelled
(for flyspell's definition of misspelled).

Repro steps:

 1. Activate flyspell so that it kicks in whenever you enter text
    mode. This I believe is the default action when you enable
    flyspell according to the basic instructions that come with it.

 2. Enter a mode which is derived from text mode, and where C-c C-c is
    used for something.

 3. Type in some text, and make sure the last word is misspelled.
    Leave the cursor just after the word in question (haven't tried
    what happens if I hit Enter before C-c C-c).

 4. Press C-c C-c

Actual result:

    A friendly beep.

Expected result:

    Whatever C-c C-c is supposed to accomplish. In mail mode, send mail.
    In vc-log-mode, submit the log entry and check in the file.

Comments:

    Step #2 can probably be generalized to cover modes not derived
    from text mode, and possibly other keystrokes in the C-c prefix
    keymap.

    Writing in a language other than English is a nice use case in
    step #3. (I haven't gotten around to figure out how to make
    Flyspell read my mind to know in what language I am going to type
    before I even know it myself.)

    For testing things without generating undesirable side effects,
    placing the cursor in various places in the text buffer and
    pressing C-h k C-c C-c is probably to be preferred. When the
    cursor is on top of a "misspelled" and thus colorized token, you
    get "C-c C-c is undefined" instead of the regular description for
    the keystroke sequence.

Sorry if this is fixed in a newer version of Emacs and/or Flyspell.
I'm on a production system so I'm running Debian stable and nothing
else here.

/* era */

-- 
Join the civilized world -- ban spam like we did! <http://www.euro.cauce.org/>
   tee -a $HOME/.signature <$HOME/.plan >http://www.iki.fi/era/index.html
Heavy spam filters in place -- try era@emacs.bugs.there.afraid.org if you do
not get a timely response from me. Phone +358-44-524-0965 (UTC+2 hours)

Re: flyspell interferes with C-c C-c (mail mode, vc-log-mode, etc)

[Richard Stallman]

Could you possibly provide one completely specific test case rather
than leaving some aspects unspecified?  Show us the text to use, say
precisely which mode to use, etc.  That is the most reliable
way to enable us to see the problem.

The Bugs section in the Emacs manual provides guidelines on this.

Re: flyspell interferes with C-c C-c (mail mode, vc-log-mode, etc)

[era eriksson]

On Tue, 19 Nov 2002 03:48:32 -0500, Richard Stallman <rms@gnu.org> wrote:
 > Could you possibly provide one completely specific test case rather
 > than leaving some aspects unspecified?  Show us the text to use, say
 > precisely which mode to use, etc.  That is the most reliable
 > way to enable us to see the problem.

I don't see which part of this is hard to come up with from the
description I gave, but here is a blow by blow transcript.

* In order to exercise vc-log-mode below, create a testing RCS file,
  in the shell.

 $ touch /tmp/testing
 $ ci -u /tmp/testing </dev/null
 /tmp/testing,v  <--  /tmp/testing
 initial revision: 1.1
 done

* Visit the file in Emacs

 $ emacs --no-site-file --no-init-file /tmp/testing
 => Note: file is write protected

* Define customary flyspell hook for text mode

 M-: (add-hook 'text-mode-hook (function flyspell-mode))
 => (flyspell-mode text-mode-hook-identify)

* Check out file for editing

 C-x v v

* Add a space

 SPC

* Prepare to check in the file

 C-x v v
 => Buffer testing.txt modified; save it? (y or n) y
 => Loading flyspell...done
 Enter a change comment. Type C-c C-c when done.

 (Other comments flashed by; see *Messages* transcript below.)

* Add a comment in the tongue of the honorable and the heroic

 Fest alla dagar

* Wait until flyspell has colorized the text

 Observe that "alla" and "dagar" are colored red to mark them as
 possible misspellings (unless of course you have ispell configured
 for Swedish by default).

 I believe I have verified that this step is necessary for the bug to
 show up. Normally I have already loaded flyspell+ispell earlier and
 am not running multiple instances of Emacs on this poor old machine,
 so this is a virtually instantaneous step, but while I was working
 out this test case, it took on the order of one second for flyspell
 to catch up and colorize the text. If I was quick enough, I could
 check in the file before the colorization had completed.

* Attempt to commit the check-in

 C-c C-c

 Get a beep back. Wonder what happened. (Tee hee, I think I know what
 happened.)

* Diagnostics: C-c C-c is "undefined" when cursor is on top of colored
  text or at the end of buffer immediately adjacent to colored text,
  while otherwise the keystroke is bound to "vc-finish-logentry" (as
  expected everywhere in the buffer).

 C-h k C-c C-c
 => C-c C-c is undefined

 M-<
 => Mark set

 C-h k C-c C-c
 => C-c C-c runs the command vc-finish-logentry
    which is an interactive compiled Lisp function in `vc'.
 (vc-finish-logentry &optional NOCOMMENT)

 Complete the operation implied by the current log entry.

 Type C-x 4 b RET to restore the other window.  M-C-v to scroll the help.

I notice that this will even reproduce under emacs -nw (or probably
other moral equivalents) although no visible colorization takes place.
(I imagine that the colorization is accomplished with an overlay which
has a keymap of its own, and that this is what causes the problem.
Could be a bug in cus-face rather than in flyspell I suppose.)

I run emacs20 and associated system software provided in the Debian
stable distribution, with the "iamerican" dictionary for ispell.

I hope this is specific enough for you. Some additional transcripts
are included below.

/* era */

*lossage* (excluding keystrokes for diagnostics):

escape : ( a d d - h o o k SPC ' t e x t - m o d e 
- h o o k SPC ( f u n c t i o n SPC f l y s p e l l 
- m o d e ) ) return C-x v v SPC C-x v v y F e s t 
SPC a l l a SPC d a g a r C-c C-c

*Messages* (excluding messages from diagnostics):

For information about the GNU Project and its goals, type C-h C-p.
Note: file is write protected
(flyspell-mode text-mode-hook-identify)
Loading vc...
Loading derived...
Loading derived...done
Loading vc...done
Checking out /tmp/testing...
Checking out /tmp/testing...done
Wrote /tmp/testing
Loading flyspell...
Loading cus-face...
Loading cus-face...done
Loading flyspell...done
Welcome to flyspell.  Use Mouse-2 to correct words.
Enter a change comment.  Type C-c C-c when done.
Starting new Ispell process...
Loading view...
Loading view...done
Type C-x 4 b RET to restore the other window.  M-C-v to scroll the help.

-- 
Join the civilized world -- ban spam like we did! <http://www.euro.cauce.org/>
   tee -a $HOME/.signature <$HOME/.plan >http://www.iki.fi/era/index.html