GNOME Bugzilla – Bug 449757
Gail code generates asserting Gtk-CRITICAL and Gtk-Warning messages with the GtkTreeView widget when Orca is running.
Last modified: 2007-08-22 20:54:55 UTC
See also Orca bug #446878 which is blocked by this bug. There are three types of message generated: 1/ cellrenderer.py:35: GtkWarning: gtk_tree_store_get_path: assertion `iter->user_data != NULL' failed gtk.main() 2/ ** (cellrenderer.py:10695): WARNING **: get_visible_column_number failed for -1 3/ Gtk-CRITICAL **: gtk_list_store_get_path: assertion `iter->stamp == GTK_LIST_STORE (tree_model)->stamp' failed Steps to reproduce (for message type #1 and #2): 1/ Start Orca 2/ Start the small Python cellrender.py example (to be attached). % python cellrenderer.py 3/ Click on the "Buy Groceries" name. 4/ Press the Esc key. Steps to reproduce (for message #3): 1/ Start Orca 2/ Start gcalctool (the GNOME calculator). 3/ Click on the View menu and make sure the Scientific radio button is set. 4/ Click on the Con button 5/ Select "Edit Constants..." 6/ Click on the value entry in row 0 (0.621). You will notice for both of the above examples, that if Orca isn't running, then the messages are not generated. I took the second example a little further: 1/ Started Orca 2/ Started gcalctool $ gdb ./gcalctool (gdb) run --g-fatal-warnings ... performed steps 3-6. Gtk-CRITICAL **: gtk_list_store_get_path: assertion `iter->stamp == GTK_LIST_STORE (tree_model)->stamp' failed aborting... Program received signal SIGABRT, Aborted.
+ Trace 142647
Thread 47453207605392 (LWP 6588)
Looks like it's a problem in gailtreeview.c
Created attachment 90388 [details] Small Python pygtk example that causes the warning messages to be generated.
Set the warning functions as breakpoint, and take a look at the trace.
By following the upper steps, we can only reproduce the following message: WARNING **: get_visible_column_number failed for -1 And we have found the reason, please see bug446878 for the detail. The most important thing is : who can tell me how to generate the other messages. (I am using Ubuntu Gutsy and latest version of a11y packages co from svn.)
> By following the upper steps, we can only reproduce the following message: > > WARNING **: get_visible_column_number failed for -1 With the latest Ubuntu Gutsy and Orca, that's the only one I can see too now. That suggests that other problems have been fixed and the other messages have gone away. If I ran the cellrenderer.py script on its own (without Orca running), the very first time I tried it, I got several occurrences of: (cellrenderer.py:23660): atk-bridge-WARNING **: failure: no device event controller found. written to the terminal window where I started it. When I tried it again on its own (after having tested it with Orca), I could not reproduce this. I will look at your comments in bug #446878, and comment there on what you've written. Thanks.
As you've probably seen, with the new comments to bug #446878, that this was indeed a bug with Orca and not atk. Closing this out as NOTABUG. Thanks again for your help with this.