Micke Andersson <[EMAIL PROTECTED]> writes: > Andrzej Adam Filip wrote: >> Anders Norrbring <[EMAIL PROTECTED]> writes: >> >> >>> Micke Andersson wrote: >>> >>>> Anders Norrbring wrote: >>>> >>>>> I have a setup with Postfix, Amavis-new and SpamAssassin on one >>>>> >>>>> >> >> 0) Micke suggested using *public IMAP folder* >> (as I understand single public folder for *all* users). >> 1) IMHO it is not a good idea to use one bayes db for all users unless >> you service small and homogeneous community. >> It is not uncommon that one person spam is another person ham. >> >> > Yes, > I would very much agree with you there, but then, Anders is using > AmavisD-new, and it does not support > personal Bayes DB's, AFAIK, AmavisD-new only supports system wide Bayes. > So then he is kind of stuck with a global handling of SPAM and HAM. > And another thing, the Q was "The easiest way"! > > However, as you said, the best way is to have personalized Bayes > training to have the best and most > accurate hits on Bayes.
IMHO it is even better to set up *individual* spam reporting to external services supported by "spamassassin -r" such as: * bulk mailing detectors (dcc/pyzor/razor2) * spamcop.net [non anonymous with requesting LART's via www *after* SMTP reporting by SA] spamcop.net LART requests may be automated by using spamcup or spamcop-ack.pl [ http://anfi.homeunix.net/perl/spamcop-ack.pl ] Passing spam protections does deserve special "reward" :-) -- [pl2en: Andrew] Andrzej Adam Filip : [EMAIL PROTECTED] : [EMAIL PROTECTED] Home site: http://anfi.homeunix.net/
