[Spambayes] SpamBayes becoming SlowBayes?

skip at pobox.com skip at pobox.com
Mon Dec 31 14:42:11 CET 2007

    Robert> I've been playing with using the SpamBayes proxy with
    Robert> Thunderbird on my Macbook Pro.  Over the last couple days I've
    Robert> noticed it become very slow, 5+ seconds to classify a single
    Robert> message, with CPU over 90%, and this with a training set of 160
    Robert> spams and 87 hams.  Is this expected performance or is there
    Robert> something going wrong?

I suspect somethings gone awry.  On my PowerBook G4 (probably at least three
times slower than your Pro), it takes about 0.8s of user+sys time to
classify a single message using sb_bnfilter.py.  I have 140 spams and 53
hams in my database.

What is your persistent_use_database config parameter set to?  What version
of SpamBayes are you using?


More information about the SpamBayes mailing list