GNOME Bugzilla – Bug 635604
Openoffice.org writer closes unexpectedly
Last modified: 2012-10-05 02:06:35 UTC
In OOo writer typing a document and then reading with arrow keys OOo closes and gives two warnings: Warning exception get_text_at_ofset() Warning exception get_run_attributes() functions. Using Orca 2.32.0 and Openoffice.org 3.2.1 shiped with Ubuntu 10.10. Since it seems that is not an Orca bug but I have filed it for evaluation.
Thanks for taking the time to report this bug. Without a stack trace from the crash it's very hard to determine what caused it. Can you get us a stack trace? Please see http://live.gnome.org/GettingTraces for more information on how to do so. Thanks in advance!
Created attachment 175106 [details] Orca debug file with possible prowide some useful information I see this problem with my Ubuntu Lucid, Orca 2.30.2, 2.32, and Orca git master version. I maked a debug.out file with Orca latest git master version. I simple launched oowriter with terminal, wrote two lines, and try close Openoffice.org Writer without saving the file (i click discard button). In debug.out file I found following traceback error message: Traceback (most recent call last):
+ Trace 224797
state = event.source.getState()
raise LookupError(e) LookupError
LookupError above while processing event: object:children-changed:add Attila
I doed my test with Libreoffice 3.3 Writer. Libreoffice 3.3 is correct closed when I choosed discard button. About dialog shows following version informations: "LibreOffice 3.3.0 OOO330m12 (Build:2) libreoffice-build 3.2.99.3" Attila
Created attachment 175185 [details] Example document causing the problem. With this document as example: 1. Place the cursor at first line "p" letter. 2. Then try to go to home / end of the document with control + home and control + end. 3. Openoofice crash and in some cases hangs up. I have to kill soffice.bin program to have control. I am using a 64 bits system. The problem seems not present with Orca 2.30.2 but 2.32.0 and later doesn't work.
* I cannot reproduce this problem with the current dev builds (OOo 3.4) * I can reproduce it with OOo 3.2. * What changed in Orca is we started listening for children-changed events in OOo so that we could announce when rows had been added to, or deleted from, a Writer table. * I can make this bug go away by removing the event listener; however, simply returning immediately in onChildrenChanged() and not processing the object event within the script, does *not* make the bug go away. * The child which was added is a dead accessible. That's right, kids, Zombies. Zombies are bad. We're probably not recognizing this as a Zombie in focus_tracking_presenter.py. I'll take this one. Even though it seems to be fixed in the dev builds, and even though this is not an Orca bug, because I hate the undead that much.
Another detail worth noting: I can reproduce this bug using the document Javier provided and without using Orca. All I have to do is launch Accerciser, enable event monitoring for object:children-changed:add events, and arrow up and down. Given this, could y'all test the latest OOo Dev (3.4) to see if the problem still persists? I'll do the same. And if the problem does still persist, I'll file the bug against OOo. Thanks.
Filed a bug against LibreOffice: https://bugs.freedesktop.org/show_bug.cgi?id=31973
In addition to not being able to reproduce this bug in OOo 3.4 dev: http://openoffice.mirrors.tds.net/pub/openoffice/extended/developer/DEV300m94/ I cannot reproduce this bug in the latest release candidate for OOo 3.3, which is available in a lot more localizations. Including Spanish, which I just tested. :-) http://openoffice.mirrors.tds.net/pub/openoffice/extended/3.3.0rc7/
While having not tried the specific document associated with this bug report I think I am still sometimes getting this bug with my openoffice 3.4 development build. However it should be noted that while I still can get openoffice to crash it does seem more reliable than it was with openoffice 3.2 (not really something hard to be, with OOo 3.2 I could crash it very quickly).
I think that between Orca fixes and AT-SPI2 fixes and LibreOffice fixes, this bug has been squashed. If you are still seeing the problem with the current Orca and current LibreOffice, please open a new bug with those details as the comments here are no longer current or relevant given how much has changed.