On 07/02/2012 2:01 PM, Philippe Michel wrote: > Running gnubg with the GDK_NATIVE_WINDOWS environment variable set to true > seems to help. >
I tried this when I first saw the bug as well. If you try this you end up with other strange things happening. For me it usually ends up with the move list pane disappearing altogether and then segfaults when I try to do other things OR not being able to continue play properly. > Googling for this finds many reports of "Ubuntu broke my application" but > what's strange is that many are rather old and relate to Ubuntu releases > where gnubg runs fine. > When I first saw this bug I thought that we had some "gtkfreeze/thaw" issues, but I looked into that and it seemed fine (Put debug output code in, and breakpoints where these occur). I was also concerned that we might have been calling GTK GUI calls from outside the main GTK thread (If you do this you have to be careful and calls some functions before hand to get things in the proper state). Again that doesn't seem to be the case. > I didn't found anything obvious about fixing that in the application > (although I don't know much about GUI programming and may not recognize > relevant information if I saw it). Still it seems likely that something > that used to work, and still does on other platforms, is not coded > defensively enough for recent Ubuntu releases. > My guess is that it is a combination of our use of GTK and interactions with the Ubuntu desktop window managers. We might have a bug (memory/buffer overrun?) as well that only manifests itself in this environment. As well, not sure you have noticed how badly our application runs on Unity with GNUBG using individual panes for output. The one thing I did not try though and would be next on my list. Debug with GDK_NATIVE_WINDOWS=1 and find out why the move pane doesn't display properly (literally disappears for me) when you click on a move that has been previously analysed. It may be that the cause of that might be related to the mouse issue ( but manifesting itself differently) - seems this scenario might be easier to track down. -- Michael Petch CApp::Sysware Consulting Ltd. OpenPGP FingerPrint=D81C 6A0D 987E 7DA5 3219 6715 466A 2ACE 5CAE 3304 _______________________________________________ Bug-gnubg mailing list [email protected] https://lists.gnu.org/mailman/listinfo/bug-gnubg
