Re: [orca-list] Please read if quitting apps makes Orca crash, hang, etc.

[Joanmarie Diggs <jdiggs igalia com>]

Hey all.

To update where things stand, the good news is that I've been able to
reproduce at least two crashers myself. I've filed these bugs against
AT-SPI2:

* https://bugzilla.gnome.org/show_bug.cgi?id=767074
* https://bugzilla.gnome.org/show_bug.cgi?id=767075

I'm hoping I'll find a way to predict what's triggering the segfault and
then not trigger it. Because I cannot reliably reproduce these crashes,
this trial-and-error prediction and work around is taking longer than I
-- and I'm guessing you -- would like. Regardless, when I've got
something new to test, I'll let you know.

In the meantime, I don't need any more debug.outs. Thanks to everyone
who sent them! If you're comfortable with gdb and manage to capture a
segfault that is not in atspi_accessible_clear_cache() or
atspi_accessible_get_parent() I would be interested in seeing it.

Stay tuned and thanks for your patience!
--joanie

On 05/31/2016 12:14 AM, Joanmarie Diggs wrote:
> Hey all.
>
> I still cannot reproduce the problems some of you have, where things
> like quitting apps causes Orca to crash, hang, etc. However, if Orca is
> running and I kill the registry daemon process manually, Orca does hang.
> So I've just committed a change to Orca master which will detect that
> condition -- if you have enabled the customization -- and shutdown when
> the registry is believed to have crashed.
>
> Yes, we of course still want to know why the registry is crashing,
> assuming that's what's happening to you. And if possible, it would be
> nice if Orca could recover on your behalf. I'll look into these things
> later today after I get some sleep. In the meantime, it would be great
> if those of you experiencing the problem could pull master and add the
> following to your orca_customizations.py:
>
> import orca.settings
> orca.settings.checkForRegistryCrash = True
>
> Then restart Orca.
>
> If, when you reproduce the problem, Orca says goodbye (hopefully) and
> quits (the process is not in the list of running processes), then the
> problem you are experiencing is almost certainly the AT-SPI2 registry
> crashing and not Orca going into some loop somewhere. Knowing whether or
> not this is the case will be helpful.
>
> Thanks!
> --joanie