Kai Schaetzl said: > Christopher M. Iarocci wrote on Fri, 20 Jun 2003 12:55:26 -0400: > > > Auto White list shows up as AWL in the report. I had this > > happen to me because I stupidly put my own domain in the whitelist by > > specifying [EMAIL PROTECTED] Of course, every spammer that spoofed an address > > at my domain got a -100 score. > > This is just expected, but I think that the auto-learn option should be > changed to not learn any whitelist or blacklist or all_spam_to mails at all. > After all, the whitelist doesn't necessarily indicate that the message in > question is ham, it just means that I want to get all mail from it (or > appearing to be from it). Also, the whitelist entry already is good enough > in detecting the message as one to let thru/not tag, so why should I feed > the Bayes learner with redundant material? > I wasn't aware about this problem until now. Consider the situation that you > have a site-wide Bayes database (as needed by spamd) and only one user is on > the all_spam_to (we don't have a machine where not at least one user is on > it). Each spam he gets is automatically fed to Bayes as ham for all users.
Not the case -- whitelist settings are ignored for auto-learning. So if the mail is judged as spam without the WL score applied, it'll be auto-learned as spam. But I agree, a better fix is to *ignore* any mails that hit a WL or BL line. There's a bug open to do this. --j. ------------------------------------------------------- This SF.Net email is sponsored by: INetU Attention Web Developers & Consultants: Become An INetU Hosting Partner. Refer Dedicated Servers. We Manage Them. You Get 10% Monthly Commission! INetU Dedicated Managed Hosting http://www.inetu.net/partner/index.php _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
