On Wed, 2008-01-30 at 11:47 -0800, gct wrote: > > OK so I've got a thread that monitors some hardware, and periodically sends > a > custom event back to the GUI to basically mirror an internal table to a > QTable I have on the GUI. I'm told this is the proper way to update GUI > elements from a thread other than the main one. My problem is, despite this > being the "correct" way, it's causing python to segfault. > > When I leave the messages enabled my GUI will run an average of 7 minutes > before I see a signal 11 and all my python processes die. Disabling the > messages (so the GUI is never updated, though the internal table is), the > run time moves all the way out to 30 minutes (at which point I manually > close the application, it does NOT segfault). > > I'm using Python 2.2.3 and PyQT 3.12 > > Has anyone run into this before, is this a known issue with customEvents > that they can segfault python?
yes, I've got exactly the same problem for several years now (across different qt/pyqt versions). I have never managed to reduce it to something feasable to be sent over to Phil, so the bug is still there :/ In my case, it's much easier to trigger (it segfaults after a couple of seconds). I've got a nasty workaround in place (I use a Python's deque to move data across threads, but the main thread as a QTimer to poll the deque and check if there's data). It doesn't work too well since the data is supposed to be read "real-time" and it's impossible to configure the timer correctly (it should go too fast and the CPU usage would top). I hope such a workaround will work with you as well. -- Giovanni Bajo _______________________________________________ PyQt mailing list PyQt@riverbankcomputing.com http://www.riverbankcomputing.com/mailman/listinfo/pyqt
