Re: [orca-list] Defunt objects causing crashes

[Francisco Javier Dorado Martínez <javier tiflolinux org>]

Hi Joanie, all

Well, with 3.16.2 I have tried to reproduce the problem for 3 times, 
only one has made it.
The other two Orca freezes for a while, then alt + tab give focus and 
recover. I have 17700.
There is no test case 100% reliable.
With master all the times I have tried the problem persists.

Anyway the problem seems to be in all versions (more or less frequent).

Would nice to know why this is happening, and if we can make any fix 
from Thunderbird script.
The problem is when Thunderbird is loading, if you load it before Orca 
then launch Orca no problem.

Many thanks,
Regards
Javier


El 05/06/15 a las 18:23, Joanmarie Diggs escribió:
> Hey Javier, all.
>
> To follow up on this: I set up a mail folder with over 32,000 messages
> and I can reproduce the problem now in which Orca does not present
> Thunderbird at all, even if you Alt+Tab out of it and back into it.
> That's the good news. The bad news is that I can also reproduce the
> exact same problem in Orca 3.16. The debug.out output is different: In
> Orca 3.16 you'll see a lot of things like this:
>
> vvvvv PROCESS OBJECT EVENT object:state-changed:focused vvvvv
> OBJECT EVENT: object:state-changed:focused             detail=(1,0,0)
> IGNORING DEFUNCT OBJECT
> TOTAL PROCESSING TIME: 0.0001
> ^^^^^ PROCESS OBJECT EVENT object:state-changed:focused ^^^^^
>
> In Orca master you'll just see an "ERROR:" line because the above output
> indicates that we took the time to queue up the event, dequeue it, and
> attempt to process it. Now we don't waste time (and performance)
> queueing, dequeuing, and attempting to process something we have
> determined was dead on arrival. But in neither case is anything spoken.
>
> I am unable to reproduce a case where Thunderbird is presented in Orca
> 3.16 but not presented in Orca master. Now that I have this mail folder
> with over 32,000 messages in it and see the problem, I'd greatly
> appreciate steps I can reproduce in which master is broken but 3.16 is
> not because if such a case still does exist, I really want to fix it.
>
> Thanks!
> --joanie
>
> On 06/05/2015 12:15 PM, Joanmarie Diggs wrote:
> > Hi Javier.
> >
> > That's helpful and disturbing at the same time. But I'll see what I can
> > do. Thanks!
> >
> > --joanie
> >
> > On 06/05/2015 12:08 PM, Francisco Javier Dorado wrote:
> > > Hi all again,
> > >
> > > The code that is causing crash reported seems to be in
> > > 6adcbd70fed3f3a5a4ad13e97457de10f5931c6e
> > >
> > > I have checked out this one and compared with previous and can reproduce
> > > the problem, however I don't have an exact test case to make this happend.
> > >
> > > Sent from 844e207713b68024616280e483d891ab15e51537
> > > Regards,
> > > Javier