On Wed, Aug 27, 2003 at 10:19:08AM -0700, Ben Gertzfield wrote: > Has all my nice Bayes training from < 2.60r2 been expired away? When I > run sa-learn on my collection of 16,000 old spam mails, it correctly > reports that it's already learned all those Message-IDs:
No, the tokens get a new timestamp during the conversion. > debug: [EMAIL PROTECTED]: already learnt > correctly, not learning twice > (many many repeats of this with different Message-IDs) the _seen database isn't changed at all. > nspam and nham are completely wrong here, as is ntokens. > > I can put up my bayes_seen and bayes_toks files, if it will help for > debugging purposes. how about just a -D output? -- Randomly Generated Tagline: "Have we had the chance to play with a blow torch Nikki?" - Mr. Wizard
pgp00000.pgp
Description: PGP signature
