Spamassassin 2.50 & SQL logging

[Jan-Peter Koopmann]

Hi,

> At the moment I'm thinking that I configure SpamAssassin to 
> use auto learning on all messages that it processes 
> (auto_learn 1 in spam.assassin.prefs.conf), and train it 
> using false positives and false negatives by getting my users 
> to use the Outlook re-send feature to send the messages to a 
> 'not-spam', 'spam' & 'ham' mailbox held on the linux box and 
> then having a cron job that runs 'sa-learn' across the 
> relevant mailboxes each day, but I was wondering what other 
> people are doing or thinking of doing in this regard??

I like the idea. Just a few questions: 

- Does the SpamAssassin bayes mechanism "use" the entire header
information especially the from/to part? How about the envelope?
- How can I convince Exchange/Outlook to really bounce a mail? The
re-send feature does not really help I am afraid.

Thanks,
  JP