[EMAIL PROTECTED] wrote: > I mean, I want the sa-learn examine the message before the spamassassin and > then goes to spamassassin > > > <snip> > -If the sa-learn determine it's spam the message will be deleted > But there's the problem. sa-learn doesn't determine if a message is spam or not. In fact, you have to explicitly tell it if the message is spam or not.
sa-learn isn't a bayes evaluator. It's a bayes trainer.