GNOME Bugzilla – Bug 690676
Crashes when clicking 'Refresh' after external file edit
Last modified: 2013-01-03 23:37:21 UTC
This is now the fourth of fifth time that Anjuta crashes after updating a file externally (in my case, using Glade2 for an older project). The change is correctly detected, and the warning bar appears over the editor. Clicking on 'Refresh' crashes Anjuta completely, not even saving last changes in editing, and IDE setup. This frequently means re-doing quite a bit of work. Normally, no info is shown at all. Until now, I was not able to catch any useful info, but I've been running inside gdb for a couple of hours now, and finally got this: (anjuta:7679): libanjuta-symbol-db-WARNING **: SQL execute_non_select failed: columns name, file_defined_id, file_position are not unique (anjuta:7679): libanjuta-symbol-db-WARNING **: SQL execute_non_select failed: columns name, file_defined_id, file_position are not unique [Thread 0x7fffb77fe700 (LWP 27314) exited] [Thread 0x7fffd227a700 (LWP 27323) exited] [Thread 0x7fffe58fa700 (LWP 27322) exited] [Thread 0x7fffb6ffd700 (LWP 27313) exited] [New Thread 0x7fffb6ffd700 (LWP 28803)] [New Thread 0x7fffe58fa700 (LWP 28812)] [New Thread 0x7fffd227a700 (LWP 28815)] [Thread 0x7fffd227a700 (LWP 28815) exited] [Thread 0x7fffe58fa700 (LWP 28812) exited] [Thread 0x7fffb6ffd700 (LWP 28803) exited] Program received signal SIGSEGV, Segmentation fault. _gtk_source_buffer_source_mark_next (buffer=0x64026a0, mark=mark@entry= 0x2353e00, category=category@entry=0x0) at gtksourcebuffer.c:1865 1865 gtksourcebuffer.c: No such file or directory. Interesting is that the SQL warnings are back. I'll check later if I made some mistake applying the patch, but I would not expect any error... This is the backtrace:
+ Trace 231317
I have the impression that these crashes only happen when in a debugging session in Anjuta. I took care this didn't happen, and had no more crashes, while I had three while debugging.
I looked at the code but it seems to be a prior memory corruption so I am unsure this is really related to the sourceview plugin.
I had no crashes at all while using Scintilla for a week, and sent back to GtkSourceview because of the impossibility of setting the tab-width. I've used it for less than 24 hours, and got another crash: (anjuta:15716): GtkSourceView-CRITICAL **: modified_changed_handler: assertion `action != NULL' failed [New Thread 0x7fffd1a78700 (LWP 4630)] Program received signal SIGSEGV, Segmentation fault. _gtk_source_buffer_source_mark_next (buffer=0x20a18f0, mark=mark@entry= 0x3a05400, category=category@entry=0x0) at gtksourcebuffer.c:1865 1865 while (mark != g_array_index (buffer->priv->source_marks, GtkSourceMark *, idx)) The backtrace is exactly the same as the previous one. To be sure, this is a freshly compiled and installed GtkSourceView.
The code paths goes away with the patch in bug 691090 so I assume that should also fix the problem as it avoids any idle handlers that might cause timing issues. *** This bug has been marked as a duplicate of bug 691090 ***