| === rob_w_away is now known as rob_w | ||
| === MacSlow is now known as MacSlow|lunch | ||
| === MacSlow|lunch is now known as MacSlow | ||
| bregma | investigating lp:827958 | 13:54 |
|---|---|---|
| cnd | bregma, which is that? | 14:03 |
| bregma | libgrip crash at gdk_x11_window_get_xid() in eog | 14:05 |
| bregma | looks like a latent bug in eog | 14:05 |
| bregma | can't be sure 'til I've tracked it down | 14:06 |
| cnd | yeah | 14:07 |
| cnd | bregma, seb128 is hounding me about it | 14:07 |
| cnd | I asked Satoris to take a look | 14:07 |
| bregma | saw that, that's why I'm looking at it | 14:07 |
| cnd | cool | 14:07 |
| Satoris | I think a good step would be to reproduce it on a build that does not have the utouch patch. | 14:09 |
| bregma | indeed | 14:10 |
| cnd | it would be great if we could do that :) | 14:10 |
| bregma | I'm working on that now: I can reproduce it with the current binary | 14:10 |
| * cnd is selfish | 14:10 | |
| cnd | oh, that's good | 14:10 |
| cnd | I didn't know how hard it would be to reproduce | 14:10 |
| bregma | I do note that there are many warnings emitted because some functions return NULL and there are no logic checks for tat | 14:11 |
| cnd | bregma, in our code? | 14:11 |
| bregma | no, in eog | 14:11 |
| bregma | most functions check their parameters and return erly instead of having logic that handles the error condition where it occurs | 14:12 |
| bregma | libgrip instead assumes it is handed valid data | 14:12 |
| bregma | I think this is the cause of the problem, libgrip assumes object invariants hold, eog does not and checks validity of everything all the time | 14:14 |
| bregma | different design philosophies | 14:14 |
| cnd | ahh | 14:15 |
| === rob_w is now known as rob_w_away | ||
| bregma | there is definitely a logic error in libgrip involving maze of twisty little passages, all different | 15:34 |
| === MacSlow is now known as MacSlow|break | ||
| cnd | bregma, does that mean you can't repro without the libgrip addition? | 16:40 |
| bregma | correct | 16:43 |
| cnd | ok | 16:43 |
| bregma | it looks like a widget and its toplevel get out of synch somehow during a convoluted sequence of signals | 16:44 |
| bregma | it's a matter of making sure everything is lined up nicely before trying to get actual X11 info | 16:44 |
| bregma | it is indeed a little maze of twisty passages, all different | 16:45 |
| === MacSlow|break is now known as MacSlow | ||
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!