Serious bug#1493

Serious bug#1493

[Chong Yidong]

If anyone has access to a remote X display, could you try to debug
#1493?  This is a crash in make-frame-on-display for remote displays,
and it's probably the most serious bug in the tree right now.

http://emacsbugs.donarmstrong.com/cgi-bin/bugreport.cgi?bug=1493

Thanks.

Re: Serious bug#1493

[Dan Nicolaescu]

Chong Yidong <cyd@stupidchicken.com> writes:

  > If anyone has access to a remote X display, could you try to debug
  > #1493?  This is a crash in make-frame-on-display for remote displays,
  > and it's probably the most serious bug in the tree right now.
  > 
  > http://emacsbugs.donarmstrong.com/cgi-bin/bugreport.cgi?bug=1493

I can't reproduce it.

From the backtrace it seems that it's using qtengine:

#10 /usr/lib/libX11.so.6(_XReply+0x243) [0x7fea4c616673]                          
#11 /usr/lib/libX11.so.6(XGetGeometry+0x8f) [0x7fea4c5f1cff]                      
#12 /usr/lib/libgdk-x11-2.0.so.0(gdk_pixmap_foreign_new_for_display+0xa0) [0x7fea4fa3eb60]                                       #13 /usr/lib/gtk-2.0/2.10.0/engines/libqt4engine.so [0x7fea48cc979e]              
#14 /usr/lib/gtk-2.0/2.10.0/engines/libqt4engine.so [0x7fea48cc3021]              
#15 /usr/lib/libgtk-x11-2.0.so.0 [0x7fea4fd7b2fd]                                 
#16 /usr/lib/libgtk-x11-2.0.so.0 [0x7fea4fdb8908]                                 
#17 /usr/lib/libgobject-2.0.so.0(g_closure_invoke+0x16d) [0x7fea4dea825d]         
#18 /usr/lib/libgobject-2.0.so.0 [0x7fea4debdc3b]                                 

and AFAIR we've had bugs in the past that were actually due to qtengine,
and not bugs in emacs.
Maybe ask the OP to try to reproduce the problem with another theme...

Re: Serious bug#1493

[Stefan Monnier]

>> If anyone has access to a remote X display, could you try to debug
>> #1493?  This is a crash in make-frame-on-display for remote displays,
>> and it's probably the most serious bug in the tree right now.
>> http://emacsbugs.donarmstrong.com/cgi-bin/bugreport.cgi?bug=1493
> I can't reproduce it.

Neither can I BTW.

> From the backtrace it seems that it's using qtengine:
[...]
> and AFAIR we've had bugs in the past that were actually due to qtengine,
> and not bugs in emacs.
> Maybe ask the OP to try to reproduce the problem with another theme...

Now that makes a lot of sense, thanks.


        Stefan

Re: Serious bug#1493

[Chong Yidong]

Dan Nicolaescu <dann@ics.uci.edu> writes:

> From the backtrace it seems that it's using qtengine:
> and AFAIR we've had bugs in the past that were actually due to qtengine,
> and not bugs in emacs.
> Maybe ask the OP to try to reproduce the problem with another theme...

OK, that makes sense.  I'll downgrade the severity to normal, then.

bug#1493: Serious bug#1493

[Chong Yidong]

Dan Nicolaescu <dann@ics.uci.edu> writes:

> I can't reproduce it.
>
> From the backtrace it seems that it's using qtengine and AFAIR we've
> had bugs in the past that were actually due to qtengine, and not bugs
> in emacs.  Maybe ask the OP to try to reproduce the problem with
> another theme...

Hi Oscar,

Can you reproduce the bug you reported without qtengine?

bug#1493: Serious bug#1493

[Oscar Fuentes]

Chong Yidong <cyd@stupidchicken.com> writes:

> Dan Nicolaescu <dann@ics.uci.edu> writes:
>
>> I can't reproduce it.
>>
>> From the backtrace it seems that it's using qtengine and AFAIR we've
>> had bugs in the past that were actually due to qtengine, and not bugs
>> in emacs.  Maybe ask the OP to try to reproduce the problem with
>> another theme...
>
> Can you reproduce the bug you reported without qtengine?

I'm not sure about what qtengine is, but I'll guess:

Choosing "Use my KDE style in GTK applications" on the appearance
settings, section "GTK Styles and Fonts", make-frame-on-display produces
a segfault, as reported. Choosing "Use another style: Raleigh" works
fine, except for this message on the console:

(emacs:12482): GLib-GObject-CRITICAL **: g_object_unref: assertion `G_IS_OBJECT (object)' failed

just after `delete-frame' on the frame created with `make-frame-on-display'.

I'm working on Kubuntu 8.10 x86_64 KDE 4.2.2.

Hope this helps.

-- 
Oscar

bug#1493: Serious bug#1493

[Dan Nicolaescu]

"Oscar Fuentes" <ofv@wanadoo.es> writes:

  > Chong Yidong <cyd@stupidchicken.com> writes:
  > 
  > > Dan Nicolaescu <dann@ics.uci.edu> writes:
  > >
  > >> I can't reproduce it.
  > >>
  > >> From the backtrace it seems that it's using qtengine and AFAIR we've
  > >> had bugs in the past that were actually due to qtengine, and not bugs
  > >> in emacs.  Maybe ask the OP to try to reproduce the problem with
  > >> another theme...
  > >
  > > Can you reproduce the bug you reported without qtengine?
  > 
  > I'm not sure about what qtengine is, but I'll guess:
  > 
  > Choosing "Use my KDE style in GTK applications" on the appearance
  > settings, section "GTK Styles and Fonts", make-frame-on-display produces
  > a segfault, as reported. Choosing "Use another style: Raleigh" works
  > fine, except for this message on the console:
  > 
  > (emacs:12482): GLib-GObject-CRITICAL **: g_object_unref: assertion `G_IS_OBJECT (object)' failed
  > 
  > just after `delete-frame' on the frame created with `make-frame-on-display'.
  > 
  > I'm working on Kubuntu 8.10 x86_64 KDE 4.2.2.

This sounds very much like a very long standing Gtk+ bug: 
http://bugzilla.gnome.org/show_bug.cgi?id=85715

we have had a few bug reports in emacs about it.

You could verify that by configuring emacs with --enable-toolkit=lucid

I think we should have a big warning in NEWS pointing to that Gtk bug
and saying that if people want to connect/disconnect to X11 display,
they'd better avoid using Gtk.