On Wednesday 19 March 2008, you wrote: > On Mar 18, 2008, at 2:44 PM, Phil Thompson wrote: > > What makes you say that? The connection has nothing to do with the > > wrapper > > object of the emitter. > > The underlying Qt connection stays alive, but on the PyQt side the > connection goes down making the Qt connection just call the universal > slot object without being forwarded to python. > > > This should make no difference. > > It does. SIP handles instance methods (and lamba's AFAIR) specifically > to avoid being dependent on a temporary wrapper object. > > I spent considerable time digging into this (although without knowing > much about the C API to Python). > If you still think I'm wrong, I could try digging back and posting > some source code references. > > In any case, you should be able to reproduce the connection problem at > your side with the example Kevin posted.
I wasn't aware that Kevin had posted code that would run. He referenced David Boddie's example which works fine for me. > FYI, these are my platform specs in case this is somehow platform > dependent: > > o Mac OS X 10.5.2 > o PyQt-4.3.3, Qt-4.3.3 > o SIP 4.7.3 > o Python-2.5.1 (the built-in version) The SIP 4.7.4 ChangeLog has the following entry... 2007/12/28 12:31:03 phil Fixed a bug where PyQt connections from an object owned by C++ were lost when the Python object was garbage collected. ...which sounds familiar. Phil _______________________________________________ PyQt mailing list [email protected] http://www.riverbankcomputing.com/mailman/listinfo/pyqt
