bug#5816: Cursor moves to end of completions in icomplete-mode

bug#5816: Cursor moves to end of completions in icomplete-mode

[Juanma Barranquero]

Version: 24.0.50


Doing

   emacs -Q -f icomplete-mode
   C-h v xa

- with the emacs-23 branch:

  Describe variable: xa<(>rgs-program)  [Matched]

- with the trunk:

  Describe variable: xa(rgs-program)  [Matched]<>


where <> marks the cursor position. Bisection seems to indicate the
bug appears in revid:eliz@gnu.org-20100330091307-ier2rhqfttgi90pr
(Initial support for bidirectional editing).

    Juanma

bug#5816: Cursor moves to end of completions in icomplete-mode

[Eli Zaretskii]

> From: Juanma Barranquero <lekktu@gmail.com>
> Date: Thu, 1 Apr 2010 04:44:22 +0200
> Cc: 
> 
> Doing
> 
>    emacs -Q -f icomplete-mode
>    C-h v xa
> 
> - with the emacs-23 branch:
> 
>   Describe variable: xa<(>rgs-program)  [Matched]
> 
> - with the trunk:
> 
>   Describe variable: xa(rgs-program)  [Matched]<>
> 
> 
> where <> marks the cursor position. Bisection seems to indicate the
> bug appears in revid:eliz@gnu.org-20100330091307-ier2rhqfttgi90pr
> (Initial support for bidirectional editing).

Yes, that figures.  The function that positions the cursor is the only
one that was completely rewritten for the bidi support, because the
original code was designed around the assumption that display is
unidirectional.  All the other bidi changes are additions to the
existing code, and are not get executed except for buffers where the
bidi display option was turned on.

Thanks for the report and the test case, I will look into this ASAP.

bug#5816: Cursor moves to end of completions in icomplete-mode

[Eli Zaretskii]

> From: Juanma Barranquero <lekktu@gmail.com>
> Date: Thu, 1 Apr 2010 04:44:22 +0200
> 
> Doing
> 
>    emacs -Q -f icomplete-mode
>    C-h v xa
> 
> - with the emacs-23 branch:
> 
>   Describe variable: xa<(>rgs-program)  [Matched]
> 
> - with the trunk:
> 
>   Describe variable: xa(rgs-program)  [Matched]<>

Actually, the problem was visible even with the standard completion,
not only with icomplete-mode.

I think I fixed it.  Please try again, and if there are still
problems, please reopen the bug.

Re: bug#5816: Cursor moves to end of completions in icomplete-mode

[Eli Zaretskii]

> From: Juanma Barranquero <lekktu@gmail.com>
> Date: Thu, 1 Apr 2010 16:03:42 +0200
> Cc: 5816-done@debbugs.gnu.org
> 
> On Thu, Apr 1, 2010 at 14:45, Eli Zaretskii <eliz@gnu.org> wrote:
> 
> > Actually, the problem was visible even with the standard completion,
> > not only with icomplete-mode.
> 
> I rarely use standard completion, so I didn't notice.

I was just surprised that no one else did.  I guess none of the users
of the trunk version really look at these responses, or maybe they
rarely type TAB more than once.