https://issues.apache.org/SpamAssassin/show_bug.cgi?id=7127
--- Comment #28 from RW <[email protected]> --- (In reply to Reindl Harald from comment #27) > easy proveable by just copy a recent trained message and only add a space to > a "bayes_ignore_header" and it is *not* learned again with --no-sync and > using jouranl as well as not removed by fdupes > > 11-02-2015 01:05:49: Verarbeite SPAM Samples > Learned tokens from 0 message(s) (3 message(s) examined) As I said it only uses the date and part of the body, so that's expected. At the start of this thread you reported that your bayes_seen file was badly corrupted. You can't trust anything you've observed with that file as being representative of normal operation. -- You are receiving this mail because: You are the assignee for the bug.
