On Sun, 2004-03-14 at 21:17, John Lagrue wrote: > Charles Howse wrote: > > That won't work. Runnng sa-learn on indivual mailboxes creates (as far > as I understand it) a set of bayesian rules in the .spamassassin > directory for that user. But as spamd is running before each user's mail > is delivered then those rules will never be applied. > > I am trying to find out how to create system-wide learned rules for SA. Since reading this thread, I've started using sa-learn regularly. I run spamd (as root), so I have made sure I run all my sa-learn invocations as root, and the bayes files are being updated in ~root/.spamassassin/bayes_* It's not clear to me from the spamd manpage if spamd it will use these files or not. If it does, then my procedure should be fine. Can anyone confirm if spamd run as root does use the .spamassassin file's in roots homedir when it starts up? Thanks, Ben