SpamAssassin preferences for every domain

[Felix Schwarz]

Hi Julian,

Julian Field wrote:
> Amavis (in all 4 split versions) runs at delivery time, so all the
> resolution email address ---> user name has been done by the MTA.
> MailScanner fits in before this resolution has been done, so would have
> to do all the resolution itself, which is "hard". It varies hugely
> between different MTAs and is impractical to implement in MailScanner.

Different SpamAssassin preferences would be apply on a from/envelope
base (not on a "real user" base).

> You can change the threshold scores, delivery actions, and black+white
> lists per user and per domain within MailScanner anyway. If you take a
> look at the Bayes scores in the most recent SA release, you will see
> they don't actually have a huge effect any more anyway, as the system
> has been largely defeated by spammers.

I'm quite suprised to see that Bayes is not very useful to many
people. I searched my private mailbox and found that most of my spam
mail are tagged with Bayes_99 (403/435 spam mails within the last two
days) or at least BAYES_9x (426/435 mails).
Because it worked so well I raised the score for Bayes_99 to 4.00 some
months ago (1 or 2 two false positives within the last month).

I would like to know how useful is Bayes to you all? Is my bayes
filter only exceptionally well trained?

--
Felix

------------------------ MailScanner list ------------------------
To unsubscribe, email jiscmail at jiscmail.ac.uk with the words:
'leave mailscanner' in the body of the email.
Before posting, read the MAQ (http://www.mailscanner.biz/maq/) and
the archives (http://www.jiscmail.ac.uk/lists/mailscanner.html).

Support MailScanner development - buy the book off the website!