On Fri, Feb 04, 2005 at 11:37:22PM -0000, Kyle Silfer wrote: > command: > su spamfilter -D -c '/usr/bin/sa-learn --spam ./unham/1.'
I don't know what -D passed to su will do. > So it's the same message ID, but the journal line indicates different entry > counts. What does that mean? Journal counts are simply the number of tokens in the journal being synced. Presumably you have filtering going on in the background, so the journal contents will be changing. > When I attempt to forget the message, here are the results: > debug: forget: message [EMAIL PROTECTED] not learnt, ignored > > Where the databases and message ID are the same, why isn't sa-learn learning > anything? What do the different journal entry counts mean? Hrm. The obvious thing is that when learning, bayes_seen isn't being updated, and for some reason there isn't an error being reported. My first thought is an issue with berkeley db. If you run db_verify on the bayes_* files, are any issues found? -- Randomly Generated Tagline: "Socialism and Nazism is the same." - Adolf Hitler
pgpUTnE0tUkgE.pgp
Description: PGP signature
