I can confirm all of Dave's remarks. I too didn't do a relearn, and I too am seeing sa-learn telling me that all the spam I have archived is already known. I went from over 400 spam to 0 after the update to 2.52.
I get the same symptoms on 2.60-cvs. Spam and ham counts have gone down considerably in the last few days. Spam count is currently 8, which is several times less the amount that gets autolearned daily. Ham count has gone down from thousands to less than 200. I don't store old spams, so re-gathering enough corpus to re-enable Bayesian filtering will take a few days. Using sa-learn to train the DB using old ham doesn't help, either.
-- Mika Hirvonen <[EMAIL PROTECTED]> http://nightwatch.mine.nu/
------------------------------------------------------- This SF.net email is sponsored by: The Definitive IT and Networking Event. Be There! NetWorld+Interop Las Vegas 2003 -- Register today! http://ads.sourceforge.net/cgi-bin/redirect.pl?keyn0001en _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
