I am a little confused as to what triggers the autolearn function.
For autolearning purposes, a second score for the email is calculated. This score is calculated as if bayes were disabled. ie: bayes rules don't count, and the appropriate non-bayes scoreset (set 0 or 1) is used.
If this score is below bayes_auto_learn_threshold_nonspam (defaults to 0.1 in SA 2.60) the message gets autolearned as ham
If the score is above bayes_auto_learn_threshold_spam (defaults to 12.0 in SA 2.60) the message gets autolearned as spam.
It's a little surprising that you never auto learn messages.. that makes me wonder if you don't have db_file installed or some other thing which is preventing bayes from being usable.
You might try running sa-learn with debugging enabled on a message or two and see if it complains about DB_File.
------------------------------------------------------- This SF.Net email sponsored by: ApacheCon 2003, 16-19 November in Las Vegas. Learn firsthand the latest developments in Apache, PHP, Perl, XML, Java, MySQL, WebDAV, and more! http://www.apachecon.com/ _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
