In a recent message here, I mentioned that I filter email in the following order:
> 1. My MTA's own, built-in validity checks (I use Courier). > > 2. My own home-grown filters installed as part of the "courierfilter" > mechanism. These reject messages based on their headers, before > they are accepted by the SMTP dialog. > > 3. My own home-grown maildrop filters that check complete messages > after they are accepted by the SMTP dialog. > > 4. SpamAssassin filtering. > > 5. TMDA's non-challenge-response filters reject appropriate messages. > > 6. TMDA's challenge-response mechanism. What I'd like to do is to take the messages that TMDA ultimately rejects in step 6 (i.e., those that are never confirmed), and feed them as "spam" to SpamAssassin's "learn" function, so that step 4 will be that much more efficient in the future. I know how to feed messages to "sa-learn", but what I don't know is how to make that happen automatically, at the point where TMDA decides that it's time to remove unconfirmed messages from the pending queue. Any ideas? Thanks in advance. -- Lloyd Zusman [EMAIL PROTECTED] God bless you. _____________________________________________ tmda-users mailing list (tmda-users@tmda.net) http://tmda.net/lists/listinfo/tmda-users
