2008/3/10, Maurice Kalinowski <haiku@xxxxxxxxxxxxx>: > Hey, > > we have had a lot of discussions about problems regarding a gcc4 build. > I for myself prefer to use that one and as you know, I currently try to > get Cortex a little bit more stable. > > Some days ago I tried to investigate a crash of Cortex and after some > time I recognized that this only happens on a gcc4 compiled system. To > be honest the backtrace does not help me at all. > > When you try to close the application the call stack is the following: > > BLooper::RemoveHandler() > BLooper::~BLooper() > BWindow::~BWindow() > RouteWindow::~RouteWindow() > BLooper::_task0_() > thread_entry(). > > So it ends up in libbe somewhere, but what is even more confusing, in a > debug build it stops in line 349 of Looper.cpp (to be a little bit more > precise). At that point we already have verified that the handler should > be ok, and I guess BList does not behave differently on these system. > > Anyone have an idea? Especially why this only occurs on gcc4? No, but I've seen similar things myself. A longshot but maybe it has something to do with virtual / non virtual destructors and how the different gcc treat them? -- Fredrik Holmqvist Chaordic: things that thrive on the edge of chaos with just enough order to give them pattern, but not so much to slow their adaptation and learning.