Python bindings trashing the stack

[John (J5) Palmieri]

I have been tracking down a stack corruption in the DBus Python
bindings.  I have been working on this since Friday when I realized that
Hal was throwing up errors and not noticing changes when I plugged in my
USB key.  The real peta is that the errors don't show up when I compile
and install by hand (./configure ..., make, make install).  They do
however show up when packaged by RPM.  Really weird is that if I take
the build done by rpm do a make clean and ./configure with the options
taken from config.log, things work again.

Wading through tones of Pyrex generated bindings made me notice that in
places we are casting python objects to dbus objects.  I am in the
process of trying to root this out and see if that improves the
situation.  In any case the bindings need a bit of a cleanup.  Just
wanted to let the list know and see if anyone else has any insight into
why things get totally trashed only when building from RPM.  The source
RPM's can be gotten from
http://people.redhat.com/johnp/redhat/experimental/utopia/SRPMS/ if
anyone wants to take a look.

-- 
John (J5) Palmieri
Associate Software Engineer
Desktop Group
Red Hat, Inc.
Blog: http://martianrock.com