[Mimedefang] How can mimedefang usespamc/spamdfuncitonsofspamassassin?

Matthew.van.Eerde at hbinc.com Matthew.van.Eerde at hbinc.com
Tue Apr 12 15:28:20 EDT 2005


David F. Skoll wrote:
>> Not necessarily;  see stream_by_recipient().
> 
>> On the other hand, that *does* eliminate any of the performance gains
>> you might make by calling SA at the MTA layer once per message,
>> rather than once per recipient.
> 
> If you can group recipients into "equivalence classes" that have the
> same SA settings, you can get away with one scan per equivalence class
> rather than one per recipient.  This is a fair bit of work.

And you end up with per-equivalence-class Bayes rather than per-user Bayes...

Perhaps something to the effect of (in filter_end)

for my $recipient (@Recipients)
{	# figure out the user associated w/ $recipient
	# make a user-specific spamassassin call
}

# figure out what to do with the message...
# if ALL spam scores are > 20, for example, maybe action_bounce...
# otherwise add user-specific headers
# X-Spam-Score-joe-at-example-dot-com: 15
# X-Spam-Score-mary-at-example-dot-com: 13

Indeed, non-trivial, and expensive.  As opposed to procmail, which is trivial-er, though perhaps more expensive. :)

Matthew.van.Eerde (at) hbinc.com                 805.964.4554 x902
Hispanic Business Inc./HireDiversity.com         Software Engineer
perl -e"map{y/a-z/l-za-k/;print}shift" "Jjhi pcdiwtg Ptga wprztg," 




More information about the MIMEDefang mailing list