GNOME Bugzilla – Bug 353130
crash in Power Management: The Information window w...
Last modified: 2007-01-29 07:02:58 UTC
Version: @VERSION@ What were you doing when the application crashed? The Information window was open, and I right-clicked the notification icon and selected "Preferences" and it crashed Distribution: Ubuntu 6.06 (dapper) Gnome Release: 2.15.92 2006-08-26 (GARNOME) BugBuddy Version: 2.15.92 Memory status: size: 123604992 vsize: 123604992 resident: 19718144 share: 13168640 rss: 19718144 rss_rlim: -1 CPU usage: start_time: 1156706132 rtime: 68 utime: 56 stime: 12 cutime:0 cstime: 0 timeout: 0 it_real_value: 0 frequency: 100 Backtrace was generated from '/opt/garnome-2.15.x/bin/gnome-power-preferences' Using host libthread_db library "/lib/libthread_db.so.1". [Thread debugging using libthread_db enabled] [New Thread 46912587976336 (LWP 12553)] 0x00002aaaafc5c094 in waitpid () from /lib/libpthread.so.0
+ Trace 71080
Thread 1 (Thread 46912587976336 (LWP 12553))
I can't reproduce. Is this repeatable? Thanks.
Ahh, looking at it more closely now, I bet it is the same as (or similar to) bug #353138 that I filed yesterday against Epiphany. That one was marked as a dup of a gail bug, so it only shows up when accessibility is enabled. Can you reproduce it with accessibility enabled? (I'm not at my computer right now so I can't test). In any case, it's probably not a g-p-m bug.
I believe this to be duplicate of bug 345074. *** This bug has been marked as a duplicate of 345074 ***
that bug says that the patch to fix the issue was in GARNOME for 2.15.91. Then it looks like the patch was removed for 2.15.92 because the fix was committed upstream. So according to 345074, that bug shouldn't be present in garnome 2.15.92, which is where I experienced this crash. So something seems slightly wrong.
Jonner, you are totally right that bug 345074 is supposed to be fixed in gail 1.9.2, as shipped with GARNOME 2.15.92. Good catch. Having a breif look at the code, the patch and bug 341853 (which also has been marked as a duplicate), both these duplicates may actually not be fixed by the patch... Trying to summon Bill or chpe to sort this out...
OK, talked to Carlos (reporter of bug 345074 and author of the patch), who is positive bug 341853 is a duplicate of his bug. Now we got this crash -- which according to the assumption above can not be a duplicate as well, given that the patch actually fixed the issue. ;) REOPENing. Moving to atk/gail. Adding chpe and Bill to Cc. Comments?
I've been investigating the source of the bug, it seems to be due to an early initialization of the a11y code in the notebook by libglade when it has only 1 page (creating its GailNotebookPage); when the rest of the pages are added, GailNotebook isn't notified (and doesn't create the respective GailNotebookPages). Then, when g-p-m removes any page (in my case the UPS one), no GailNotebookPage is found. Note that all this happens before entering the main loop, at some point inside it atk_object_ref_accessible_child() is called for each child, which ends up creating the rest of the GailNotebookPages. If I may propose a solution, I'd connect to the ::page-added signal (in GtkNotebook since 2.10) and create the respective GailNotebookPage in its handler, I'll make a patch tomorrow if I've got some time.
Created attachment 72222 [details] [review] patch to fix the issue, implements the proposed solution
Thanks Carlos! Please commit ASAP or let me know and I'll commit tomorrow (wed 13-sept)
I committed this today. Thanks again Carlos!
*** Bug 352042 has been marked as a duplicate of this bug. ***