Highlighting bug

Highlighting bug

[Jorge A. Alfaro-Murillo]

Hi!

After the merge from emacs-26, the highlighting of a region 
stopped working for me after setting the mark or selecting a 
region with the mouse.

I bisected the problem and it was introduced in commit 
f6818e761eaafe095e07249180dc8f9a329f1473

Since this seems to be related to GTK, for the record, I have 
libgtk2.0-0 version 2.24.30-4. But it might be relevant to let you 
know that I use openbox, run as a standalone session.

Best,
-- 
Jorge.

Re: Highlighting bug

[Kaushal Modi]

[-- Attachment #1: Type: text/plain, Size: 1631 bytes --]

On Wed, Sep 27, 2017 at 10:46 AM Jorge A. Alfaro-Murillo <
jorge.alfaro-murillo@yale.edu> wrote:

> Hi!
>
> After the merge from emacs-26, the highlighting of a region
> stopped working for me after setting the mark or selecting a
> region with the mouse.
>
> I bisected the problem and it was introduced in commit
> f6818e761eaafe095e07249180dc8f9a329f1473
>
> Since this seems to be related to GTK, for the record, I have
> libgtk2.0-0 version 2.24.30-4. But it might be relevant to let you
> know that I use openbox, run as a standalone session.
>

Thanks for posting this. I have seen this bug, but haven't yet figured out
how to consistently reproduce it.

Symptoms:

C-SPC M-f

(nothing happens! the text does not get selected (or doesn't seem to))

Then I do M-x revert-buffer, and C-SPC M-f starts working again.

I have faced this infrequently, and don't yet have a clean recipe to
reproduce it.

My Emacs version:

Emacs version: GNU Emacs 26.0.60 (build 7, x86_64-pc-linux-gnu, GTK+
Version 2.24.23)
 of 2017-09-25, built using commit 827db6b559100153fd7dcab1ecdabd9233e906ab.

./configure options:
  --with-modules --prefix=/home/kmodi/usr_local/apps/6/emacs/emacs-26
'--program-transform-name=s/^ctags$/ctags_emacs/'
'CPPFLAGS=-I/home/kmodi/usr_local/6/include -I/usr/include/freetype2
-I/usr/include' 'CFLAGS=-ggdb3 -O0' 'CXXFLAGS=-ggdb3 -O0'
'LDFLAGS=-L/home/kmodi/usr_local/6/lib -L/home/kmodi/usr_local/6/lib64
-ggdb3'

Features:
  XPM JPEG TIFF GIF PNG RSVG IMAGEMAGICK SOUND GPM DBUS GSETTINGS NOTIFY
ACL LIBSELINUX GNUTLS LIBXML2 FREETYPE LIBOTF XFT ZLIB TOOLKIT_SCROLL_BARS
GTK2 X11 MODULES

-- 

Kaushal Modi

[-- Attachment #2: Type: text/html, Size: 2364 bytes --]

Re: Highlighting bug

[Jorge A. Alfaro-Murillo]

Thanks, Kaushal.

Kaushal Modi writes:

> I have seen this bug, but haven't yet figured out how to 
> consistently reproduce it. 
> 
> Symptoms: 
> 
> C-SPC M-f 
> 
> (nothing happens! the text does not get selected (or doesn't 
> seem to))

For the record, selecting the region does work for me. So in the 
case that you mention (C-SPC M-f) the point and the mark are at 
the right places (after the word and before the word, 
respectively). The problem for me is just that it is not 
highlighted properly.

Also, when running emacs with -nw, there is no highlighting 
problem.

-- 
Jorge.

Re: Highlighting bug

[Philipp Stephani]

[-- Attachment #1: Type: text/plain, Size: 627 bytes --]

Jorge A. Alfaro-Murillo <jorge.alfaro-murillo@yale.edu> schrieb am Mi., 27.
Sep. 2017 um 16:46 Uhr:

> Hi!
>
> After the merge from emacs-26, the highlighting of a region
> stopped working for me after setting the mark or selecting a
> region with the mouse.
>
> I bisected the problem and it was introduced in commit
> f6818e761eaafe095e07249180dc8f9a329f1473
>
> Since this seems to be related to GTK, for the record, I have
> libgtk2.0-0 version 2.24.30-4. But it might be relevant to let you
> know that I use openbox, run as a standalone session.
>

Hmm, interesting. The change shouldn't affect GTK 2 at all, only GTK 3.

[-- Attachment #2: Type: text/html, Size: 952 bytes --]

Re: Highlighting bug

[Jorge A. Alfaro-Murillo]

Philipp Stephani writes:

> Hmm, interesting. The change shouldn't affect GTK 2 at all, only 
> GTK 3. 

I also have libgtk-3-0 version 3.20.6-1.

-- 
Jorge.

Re: Highlighting bug

[Kaushal Modi]

[-- Attachment #1: Type: text/plain, Size: 351 bytes --]

On Thu, Sep 28, 2017 at 4:19 PM Jorge A. Alfaro-Murillo <
jorge.alfaro-murillo@yale.edu> wrote:

> Philipp Stephani writes:
>
> > Hmm, interesting. The change shouldn't affect GTK 2 at all, only
> > GTK 3.
>
> I also have libgtk-3-0 version 3.20.6-1.
>

Then may be the issue I am seeing is different.. I only have GTK2 on RHEL
6.6.
-- 

Kaushal Modi

[-- Attachment #2: Type: text/html, Size: 760 bytes --]

Re: Highlighting bug

[martin rudalics]

 > I bisected the problem and it was introduced in commit
 > f6818e761eaafe095e07249180dc8f9a329f1473
 >
 > Since this seems to be related to GTK, for the record, I have
 > libgtk2.0-0 version 2.24.30-4. But it might be relevant to let you
 > know that I use openbox, run as a standalone session.

Is GTK_CHECK_VERSION broken?  If you put a breakpoint in gtkutil.c at
the "return false" statement below does it get hit?

#if GTK_CHECK_VERSION (3, 16, 0)
   if (get_bg)
     /* gtk_style_context_get_background_color is deprecated in
        GTK+ 3.16.  New versions of GTK+ don't use the concept of a
        single background color any more, so we can't query for it.  */
     return false;
#endif

martin

Re: Highlighting bug

[Jorge A. Alfaro-Murillo]

martin rudalics writes:

> Is GTK_CHECK_VERSION broken?  If you put a breakpoint in 
> gtkutil.c at the "return false" statement below does it get hit? 
> 
> #if GTK_CHECK_VERSION (3, 16, 0) 
>   if (get_bg) 
>     /* gtk_style_context_get_background_color is deprecated in 
>        GTK+ 3.16.  New versions of GTK+ don't use the concept of 
>        a single background color any more, so we can't query for 
>        it.  */ 
>     return false; 
> #endif 
> 

I do not know C, so what exactly should I do? If I put a 
raise(SIGINT); right before the return false; and compile, then 
emacs crashes when I try to highlight a region, so I guess it does 
get there?
-- 
Jorge.

Re: Highlighting bug

[martin rudalics]

 > I do not know C, so what exactly should I do? If I put a
 > raise(SIGINT); right before the return false; and compile, then emacs
 > crashes when I try to highlight a region, so I guess it does get
 > there?

Then GTK_CHECK_VERSION seems broken indeed.  Funny.

martin

Re: Highlighting bug

[Jorge A. Alfaro-Murillo]

The following message is a courtesy copy of an article
that has been posted to gmane.emacs.devel as well.

martin rudalics writes:

>> I do not know C, so what exactly should I do? If I put a 
>> raise(SIGINT); right before the return false; and compile, then 
>> emacs crashes when I try to highlight a region, so I guess it 
>> does get there? 
> 
> Then GTK_CHECK_VERSION seems broken indeed.

But I do have a libgtk-3-0 higher than 3.16 (in fact I just 
upgraded from 3.20.6-1 to 3.22.21-1). So I should be getting into 
the "if GTK_CHECK_VERSION (3, 16, 0)", right? Or does this depend 
on the version that I am using, because I also have 
libgtk2.0-0. Should I have only one of those two installed?

Could this a problem of openbox using libgtk2.0-0 even when 
libgtk-3-0 is installed? Because when I remove from gtkutil.c the 
lines of the patch:

-#if GTK_CHECK_VERSION (3, 16, 0) if (get_bg) /* 
-  gtk_style_context_get_background_color is deprecated in 
-  GTK+ 3.16.  New versions of GTK+ don't use the concept of a 
-  single background color any more, so we can't query for it.  */ 
-  return false; 

-#if GTK_CHECK_VERSION (3, 16, 0) /* We can't get here.  */ 
-      emacs_abort (); #else 

-#endif

Then the problem goes away.

Thanks!
-- 
Jorge.

Re: Highlighting bug

[martin rudalics]

 > But I do have a libgtk-3-0 higher than 3.16 (in fact I just upgraded
 > from 3.20.6-1 to 3.22.21-1). So I should be getting into the "if
 > GTK_CHECK_VERSION (3, 16, 0)", right? Or does this depend on the
 > version that I am using, because I also have libgtk2.0-0. Should I
 > have only one of those two installed?

Then you probably built with GTK 3.22 and the bug is a consequence of
Philipp's change.  Maybe you should write a proper bug report.

martin

Re: Highlighting bug

[Jorge A. Alfaro-Murillo]

Martin Rudalics writes:

> Then you probably built with GTK 3.22 and the bug is a 
> consequence of Philipp's change. Maybe you should write a proper 
> bug report.

Thanks. Philipp already reverted the change (commit 5406be4) and 
submitted a patch. I just compiled and everything is working 
properly again.

Best,
-- 
Jorge.