[Python-modules-team] Bug#558112: Probably fixed in upstream pyqt 4.6.2

[Crestez Dan Leonard]

I can confirm this bug. It's very easy to reproduce; even from gnome.
Just apt-get install printer-applet and gdb --args
python /usr/bin/printer-applet. I get the same backtrace every time.

I believe this bug is from upstream pyqt and that it's fixed in 4.6.2
(the debian version is currently 4.6.0).

I reported a bug in anki upstream with a very similar backtrace. They're
both crashes inside python itself coming from a qt event.
Link: http://code.google.com/p/anki/issues/detail?id=1512

The author claimed it was not a bug in anki but in pyqt/mathplotlib
interaction. Here's a relevant thread from the mathplotlib list:
http://sourceforge.net/mailarchive/message.php?msg_id=4AF1FE9E.5030609%
40pythonxy.com

They claim it's a pyqt issue fixed in 4.6.2. The upstream pyqt changelog
seems to confirm this:
""" Fixed a regression in the handling of QObject.destroyed() being
connected via a proxy that meant that the proxied callable never got
invoked because the proxy was being destroyed too soon. """
http://www.riverbankcomputing.co.uk/static/Downloads/PyQt4/ChangeLog

I did not actually build pyqt 4.6.2 from source and test this theory.