python 2.3a2 Win32 expected failures in regression tests?

[Harri Pasanen]

Tim Peters <tim.one at comcast.net> wrote in message news:<mailman.1049910795.4707.python-list at python.org>...

> ... 2.3a2 is so old now that it would be better to
> test current CVS Python.

I poked around Python CVS, but I did not find any Changes.log, or
equivalent, to say what has changed since 2.3a2, or from 2.2.2 to
that. What would be the best way to find out?

I chasing an obscure bug where OmniORB's omniidl fails when I build it
using Python 2.3a2 on W2K, while on Linux it works.  On W2K it works
with Python 2.2.2.   Some of its internal import's seem to fail
silently, providing partially imported modules, but only for 2.3a2,
W2K combo.   The culprit is either me, omniorb, or python, and I'm
trying to pin it down.  I seem to recall there were changes to the
import mechanism since 2.2.2?

Regards,

Harri