[Spambayes] MAPI error prevents message scanning

[Meyer, Tony]

>> By all means, jump in!  ( you'll like Python more than Perl 
>> anyway :p ).
> I try not to be a language bigot, so you might be right.  I 
> just haven't had
> occasion thus far to get familiar with it.

I was only teasing - I only know the tiniest amount of Perl, so I
wouldn't really know.

[imapfilter explanation]
> Hmm, interesting.  I guess I assumed it worked similar to the 
> POP3 proxy.
> I'll have to look at the code...

The people requesting an IMAP solution at the time it was created were
more in favour of a filter than a proxy (similar to isbg and
SpamAssassin) so that was the solution that was made.  I do think IMAP
suits a filter more than a proxy, and it does mean that imapfilter
operates closer to the Outlook plugin style, which is generally seen as
a reasonably easy to use.

> I'd be very interested in seeing the code.

I'll try and get some of this checked in soon then.  (It works (or seems
to ;) with Mozilla Mail at the moment, but not (sigh) with OE...).

> Thanks for mentioning 'Twisted'; I'm going to learn more about it.

A warning: the API for the imap part of twisted isn't all that stable (I
think the client part is more so than the server, but I haven't used the
client stuff).  In the process of building my server, I've had to submit
various bugs/patches, some of which are in the current release, some of
which aren't yet.  If you do go the twisted way, it's definitely worth
keeping an eye on the mailing list & bug tracker.

> Rather than hack up my own Bayesian filter, I'd rather try to 
> contribute a superstructure over SpamBayes that would help make
> it a corporate solution to spam rather than a distributed (on
> the desktop) solution.

Well, there has definitely been a lot of interest in a server side
version, although nothing has eventuated yet, AFAIK (the python.org mail
being an exception, but not quite the same thing).

> Besides, it's a good excuse to learn a new language...  <grin>

Indeed!

=Tony Meyer