I have no idea about the cause, but it's happening every time at startup. I'm using xfce4-appfinder-4.10.0-1.fc17.x86_64 on Fedora 17 from nirik's repository. Attached is the report generated by the abrt tool.
Created attachment 4548 abrt.log
Created attachment 4552 appfinder-categories-nil-value.patch Hi Cristian, :#0 xfce_appfinder_category_model_icon_theme_changed (model=0x1cd6e40) at appfinder-category-model.c:591 : item = <optimized out> : li = 0x1 : idx = <optimized out> : iter = {stamp = 680912380, user_data = 0x1c31d60, user_data2 = 0x1cd59e0, user_data3 = 0x40daf9} : path = <optimized out> "li" is a GSList, and it's value is 1, that is incorrect and is the cause of the segfault. Is it possible for you to test the attached patch? It looks like an unconfigured icon theme is causing this, perhaps the icon theme is not yet initialized when appfinder starts; of course appfinder must not crash because of this. Regards, Mike
Since I'm using RPMs, it's not easy for me to apply the patch, but I've asked nirik to incorporate it in his RPMs.
nirik was kind enough to apply the patch the RPM and now I'm using it. I remember still having some issues afterwards, but right now everything looks fine. If I'll encounter another crash, I'll leave a comment here.
categories is already NULL, because the object is memset(0).
I've managed to make xfce4-appfinder crash again under similar circumstances, but I wasn't able to get a stack trace, so I'm not sure it was the same bug.
Created attachment 4573 abrt2.log abrt log after applying the patch
Added additional protects (r4244183), but it looks like the corruption is happening somewhere else, to check that you need to run appfinder in valgrind.
I have xfce4-appfinder-4.10.0-3.fc17.x86_64 [1] right now and I haven't had any issues with it lately. [1] http://repos.fedorapeople.org/repos/kevin/xfce-4.10/fedora-17/fedora-17-x86_64/xfce4-appfinder-4.10.0-3.fc17/build.log
I've upgraded to Fedora 18 in the mean time and the plugin seems to work better, but it still crashes from time to time, although under other usage scenarios.
This bug is very old and it probably is not reproducible anymore (well, it never crashed for me). If this is still an issue, please reopen.