Spambuckets, Bayes and MailScanner signatures

[Stef Morrell]

Hi all,

Having recently gotten my head around extracting RFC822 email from
exchange servers using IMAP, I'm considering setting up a spambucket, so
my users can dump false negatives - then using some kind of suitable
script to feed them into sa-learn.

Now, Bayes has already been told to ignore the X-MailScanner-Blah
headers, in the spamassassin prefs, but I'm wondering about how it will
react to being fed things like the inline anti-phishing stuff and also
the "This has been scanned by MailScanner" etc signature.

Obviously what I don't want is for Bayes to get wrong ideas from dodgy
data. GIGO :)

Do I need to somehow process those bits out in an effort to restore the
original email, or does the order in which things are done mean that
it's not terribly relevant?

Regards

Stef
Stefan Morrell          | Operations Director
Tel: 0845 3452820       | Alpha Omega Computers Ltd
Fax: 0845 3452830       | Incorporating Level 5 Internet
stef at aoc-uk.com         | stef at l5net.net