On Tue, 29 Mar 2011 19:09:08 -0700 jdow <j...@earthlink.net> wrote: > On 2011/03/29 11:30, RW wrote: > > On Tue, 29 Mar 2011 12:55:51 -0500 > > Max<mdun...@breakawaysystems.com> wrote: > > > >> Heres the output of spamassassin -D --lint: > >> > >> [29434] dbg: logger: adding facilities: all > >> [29434] dbg: logger: logging level is DBG > >> [29434] dbg: generic: SpamAssassin version > > > > Update to the current version. It's not worth giving it any more > > thought until you've done that. The rules for 3.2.5 haven't been > > worked on some time. > > Nuts, 3.2.5 works well enough. Your comment basically tells the world > that it was not the update that caused the problem. It's more like > broken training.
Spamassassin should still produce reasonable results even if BAYES isn't working properly. If an installation is reliant on BAYES then you might as well replace it with a pure statistical filter, you'll get better results with less resources. Unless the OP is holding back about lots of extra spam that *is* being caught I'd say there are wider problems beyond BAYES. Possibly the old 3.2.5 rulesets are even causing the BAYES problem via poor autolearning. My recollection is that most threads about scoring problems on 3.2.5 usually end-up with the observation that it wouldn't have happened on 3.3.x.
