Problems with emacs built against gtk3

[Ken Brown]

On 12/2/2011 5:35 AM, Yaakov (Cygwin/X) wrote:
> On Thu, 2011-12-01 at 18:04 -0500, Ken Brown wrote:
>> On 11/30/2011 9:09 AM, Ken Brown wrote:
>>> On 11/29/2011 10:51 PM, Yaakov (Cygwin/X) wrote:
>>>> This appears to be the same bug. The solution is to launch a DBus
>>>> session bus *before* starting emacs (or any other gtk3 programs for that
>>>> matter), IOW:
>>>>
>>>> $ eval `dbus-launch --sh-syntax`
>>>> $ emacs-X11&
>>>
>>> That gets rid of the warning, but emacs still dies after a few seconds
>>> (no error message, no stackdump), unless I uninstall dconf-service. I'll
>>> see if I can get more information by running emacs under gdb. I'd
>>> appreciate any suggestions you might have as to where I should look.
>>>
>> I have some further information: The problem is related to the GSettings
>> backend.  If I uninstall dconf-service and start emacs, I get a warning
>> that the GSettings `memory' backend will be used.  Emacs then works
>> fine.  If I reinstall dconf-service but set GSETTINGS_BACKEND=memory
>> before starting emacs, it again works fine.  Does this provide any clue
>> as to what the problem might be?
>
> Okay, I got it.  dconf-service needs a GVfs implementation, but the
> default provider (from the gvfs package) is currently only available in
> Ports.  That's what I get for trying to be minimalistic wrt the distro.
>
> Of course, my gvfs package requires Avahi[1], so it may be an
> interesting ITP; I'll try to do that next week, and hopefully this
> thread will help expedite the review nonetheless.
>
> In the meantime, try setting the GIO_USE_VFS environment variable to
> "local"[2], which will allow dconf-service to work despite the lack of
> gvfs.  (Why this isn't done automatically as a fallback, I have no
> idea.)

This doesn't do it.  Emacs still dies after a short time.  I don't know 
if that means that there's something else going on, but I'll retest it 
after you package gvfs.  In the meantime, I'll continue with my 
workaround of setting GSETTINGS_BACKEND=memory.

In case you (or anyone else) wants to experiment with this, you can get 
my build of the emacs-24 pretest by running

    setup.exe -K http://www.math.cornell.edu/~kbrown/kbrown.gpg

and adding http://www.math.cornell.edu/~kbrown to the list of mirrors.

Ken

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Problem reports:       http://cygwin.com/problems.html
Documentation:         http://x.cygwin.com/docs/
FAQ:                   http://x.cygwin.com/docs/faq/