[Spambayes] How to feed a spam corpus to a MTA?

[Amedee Van Gasse]

I have a question that's not directly related to Spambayes but to 
antispam in general. But I know that there are a lot of people on this 
list who could point me in the right direction.


I'm writing a paper comparing several antispam techniques. I would like 
to make some measurements too.
To do that, I'm planning on sending the same set of messages (several 
thousand, maybe lots more) trough the filters and see how they perform.

I have a practical question. It won't be a problem to feed spam to a 
filter that works on the mail itself, but how do I test with properties 
of the SMTP session, like the original MTA? The original "metadata" 
isn't saved in an email in a file. I could feed the spam corpus to a 
postfix, who will behave as a new MTA, but then it will be the same MTA 
for all messages.

Or is this just not possible? Will I have to settle with live testing 
for tests like DNSBL, greylisting,... and test the spam corpus only on 
things like spamassassin, spambayes,... ?

-- 
Amedee