So far, the best approach I have found thus far for TB is the Bayes
Filter Plugin (by Achim Winkler) plus SpamBayes. TB plus the BFP runs
at 19 megs of RAM; SpamBayes (using the same training data, more than
3,000 spam and 3,000 ham) uses just 11 megs. I used POPfile for a long
time, but it was too RAM-heavy, much too slow in startup for multiple
accounts, and generally ate up resources on my XP/800MHz/392MB machine.
--
Jonathan E. Brickman
http://joshuacorps.org
________________________________________________
Current version is 2.10.03 | 'Using TBUDL' information:
http://www.silverstones.com/thebat/TBUDLInfo.html