[Spambayes] using external DNSBL's with Outlook

[Seth Goodman]

> [Seth Goodman]
>
> > When exactly does SpamBayes get to work on a message?
>
> Heh:  nobody knows.  The Outlook rule system isn't strong enough to allow
> plugging SpamBayes in as an Outlook rule, so instead SpamBayes
> hooks Outlook
> "folder add" events.  Exactly when and why Outlook triggers one of those
> events isn't documented by Microsoft, and appears to vary wildly depending
> on Outlook version and system load.  Sometimes SpamBayes appears to see a
> message before Outlook rules run, sometimes after they run, sometimes we
> don't get notified about a new message at all, and sometimes we
> get notified
> about a new message more than once (e.g., if an Outlook rule
> moves a message
> and SpamBayes is set to watch both the Inbox and the destination
> folder, we
> *might* be told about a message twice).
>

That sounds like it was really hard to work around.  Microsoft probably
didn't document the relationship to the rule processor because they don't
really know themselves.  I bet there were too many people involved and the
code's an unholy mess.  The only way I can think of to avoid this (I'm sure
you folks have beat this one up and down) is to install yourself as a proxy
so you get the message before Outlook mangles it.  You'd have to mark up the
headers with the results and then have the Outlook rules process things
"normally".  Of course, that would make all the nice stuff that you've added
in Outlook nearly impossible.

This is a wonderful package you've put together.  I particularly appreciate
how easy it is to simultaneously retrain the classifier and move the message
to it's proper folder by just hitting a button.  That makes operation nearly
effortless.  I'll fool around with the background processing option to see
if I can get DNSBL tagged messages moved before SpamBayes classifies them.
Alternatively, is there some way that I can tell SpamBayes to tokenize the
particular X- header line that contains the DNSBL tag?  I'm not even certain
that would achieve the desired result.

Regards,

Seth Goodman

Goodman Associates, LLC