"Jason R. Mastaler" <[EMAIL PROTECTED]> writes:

> I also noticed that TMDA/Auth.py contains a nearly identical function
> (L254).  Do these fixes need to be applied there as well?

I'm tracking this down because of a question on -users.  Originally,
another poster had troubles with tmda-ofmipd and I meant to update
this and forgot.  The problem that today's poster mentioned reminded
me.  When I went back to answer his question, I discovered that he was
talking about tmda-cgi.

That's when I discovered Auth.py.  It's not used by anything but
tmda-cgi, but apparently the idea was that we would eventually move
tmda-ofmipd to use it, too.  The problem is, I'm scared of it.  It is,
to put it kindly, a work in progress.

To answer your specific question, the test for whether it needs to
define its own IMAP4_SSL class or not looks correct in Auth.py.  It
imports imaplib.py and attempts to access the IMAP4_SSL class.  Only
if it is unsuccessful does it create its own.  That seems right to
me.

Something is obviously still screwy, though, since 'remote' is being
passed to the IMAP4_SSL class's __init__.  When I had looked at
tmda-ofmipd, I believed that was because of the incorrect use of
'self.host' and 'self.port' in tmda-ofmipd's IMAP4_SSL.open().  That
bug would only manifest itself when tmda-ofmipd's class descended from
the IMAP class in Python 2.3 and *not* in previous Python versions and
the only reason it would do that was because of the bogus version
test.

Now... I'm not so sure.  Because of the way Auth.py decides whether to
create its own IMAP4_SSL class, the problem should never occur, but it
still seems to be happening, so I need to look at this a little more.

There's nothing wrong with the tmda-ofmipd fixes I checked in, but
there may be more to the problem.  I'll keep looking.


Tim

_________________________________________________
tmda-workers mailing list ([EMAIL PROTECTED])
http://tmda.net/lists/listinfo/tmda-workers

Reply via email to