> Because autolearning as spam is not as simple as 28.2 >10. > > 1) Autolearning is NOT determined by the normal message score. It's > determined by the score it would have gotten if bayes was disabled. This > includes changing the scoreset, so you have to re-add everything. The > difference can be really huge.
Oh .. ok .. thanks. Now it's clear. I think I'm ready to update to 3.1.0, but last question: I want to update from DB_File 1.75 to 1.811. Will there be any problem with the current bayes db? Will I have to backup and restore it manually? I don't want to loose the db ... Thx Michael
