This is just an quick update, as I'm a bit confused right now...
Can you please confirm me that using the testing/unstable package recompiled for etch is not affected by the bug?I'll try. Seems I missed to include support for mx in my build so it takes some changes to the program, but it shouldn't be to hard to fix. I'll get back to you soon.
I installed the new (backported) version of psyco and modified the program to user regular python datetime instead of mx.DateTime. Tried it a couple of times and it worked. I then purged the new 2.0.7 version and reinstalled the default etch version (2.0.5.1), run the same (new) program and it still didn't fail!! Finaly I run the original version, the one using mx.DateTime, and at least that version still fails as before. So I'm begining to wonder whatever something else is wrong. Is there a way to guild the 2.0.7 version with support for mx? That way I could run the _exact same program_ that failed before... // Emil