If you upgrade to 2.60 you need to set your autolearn ham score to 0, instead of the default of -2 in 2.55.
By default shouldn't the auto-learn threshold for 2.60 be set to 0? (not at a shell prompt so can't look right now.) If not I think it should be, otherwise it won't learn ham. --Luke As "Matt Kettler" <[EMAIL PROTECTED]> said: > There's never been any sort of minimum. > > However, in 2.60-rc3 there are very few negative scoring rules. It would > be > considerably less common for a message to get a sub-zero score.. Check a > message from someone that's whitelisted, or has a Habeas SWE header in it. > > 2.55 had about 64 rules with negative scores. 2.60-rc3 has 13. > > $ grep "score" 50_scores.cf | grep - | grep -v \# > > score RCVD_IN_BSP_OTHER -0.1 -0.1 > score RCVD_IN_BSP_TRUSTED -4.3 -4.3 > score USER_IN_WHITELIST -100.000 > score USER_IN_DEF_WHITELIST -15.000 > score USER_IN_WHITELIST_TO -6.000 > score USER_IN_MORE_SPAM_TO -20.000 > score USER_IN_ALL_SPAM_TO -100.000 > score HABEAS_SWE -8.0 > score BAYES_00 -4.901 -4.900 > score BAYES_01 -0.600 -1.524 > score BAYES_10 -0.734 -0.908 > score BAYES_20 -0.127 -1.428 > score BAYES_30 -0.349 -0.904 > > > > > ------------------------------------------------------- > This sf.net email is sponsored by:ThinkGeek > Welcome to geek heaven. > http://thinkgeek.com/sf > _______________________________________________ > Spamassassin-talk mailing list > [EMAIL PROTECTED] > https://lists.sourceforge.net/lists/listinfo/spamassassin-talk > ------------------------------------------------------- This sf.net email is sponsored by:ThinkGeek Welcome to geek heaven. http://thinkgeek.com/sf _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
