On Wednesday 14 of January 2015 15:49:20 Bastien Nocera wrote: > On Wed, 2015-01-14 at 15:33 +0100, Jakub Filak wrote: > > On Wednesday 14 of January 2015 14:54:51 Bastien Nocera wrote: > > > On Wed, 2015-01-14 at 14:48 +0100, Jakub Filak wrote: > > > > On Wednesday 14 of January 2015 14:19:44 Bastien Nocera wrote: > > > > > On Wed, 2015-01-14 at 14:09 +0100, Jakub Filak wrote: > > > > > > On Wednesday 14 of January 2015 12:06:21 Bastien Nocera wrote: > > > > > > > None of the actions we're supposed to be taking after the main > > > > > > > loop > > > > > > > should be necessary at that point in time. Let GTK+ handle that > > > > > > > as > > > > > > > it > > > > > > > should by exiting if the SIGTERM signal is sent. > > > > > > > > > > > > Can you please point me to some documentation dealing with the > > > > > > signals > > > > > > and > > > > > > GTK+? > > > > > > > > > > There's no documentation, there's source code though. It doesn't > > > > > handle > > > > > SIGTERM especially, so would just "terminate" if sent SIGTERM. > > > > > > > > > > > I applied your patch, started abrt-applet, sent SIGTERM to it and > > > > > > the > > > > > > process was "Terminated" but it should gracefully exit. > > > > > > > > > > > > I use SIGTERM to test the applet's behaviour when Desktop is > > > > > > terminating. > > > > > > > > > > What would abrt-applet do on exit that's required to be done on > > > > > exit? > > > > > > > > abrt-applet notifies a user about problems that occurred while his > > > > system > > > > was not running a desktop session. The applet has a list where it > > > > stores > > > > the seen problem IDs and the applet sends a notification for all > > > > not-yet-seen problems at start-up. The seen list must be updated > > > > before > > > > the applet exits because of unclear legacy reasons I am not able to > > > > recollect and are possibly gone, but I'm not sure and I'd rather leave > > > > it > > > > as it is. > > > > > > Keeping a list of seen items seems like the sort of thing that the > > > application crashing/going away suddenly shouldn't affect. > > > > See my comment : > > https://github.com/abrt/abrt/blob/master/src/applet/applet.c#L1828 > > > > > Why don't we > > > remove it now, and see whether it causes any problems instead? > > > > OK, is there any other reason to remove it except you consider that code > > as > > pointless? > > > > Please don't get me wrong. I really appreciate your efforts. I just have a > > different style of work :) > > So we write code for the 0.3% cases? Yes, we do have different styles of > work indeed...
:D I just didn't remove that code, it was there before we improved the applet to be able to remember 99.7% of detected problems without the signal handler. I just don't touch/remove stuff that works unless it is really necessary. Jakub
