Personally $me is using simple bash-script, like
...
#!/bin/bash
echo "Learning SPAM..."
sa-learn --progress --spam /opt/system/mail/spamfilter/spam/
rm -fr /opt/system/mail/spamfilter/spam/*
echo "Learning HAM..."
sa-learn --progress --ham /opt/system/mail/spamfilter/ham/
rm -fr /opt/system/mail/spamfilter/ham/*
...
I am using my own delivery system, and users are able to forward mails to
addresses like getspam@ and
[EMAIL PROTECTED] You can configure procmail to do the same thing. You may also
use crontab to periodically run
sa-learn script. In my opinion it isn't good to run sa-learn exactly in
procmail, because it may cause high CPU usage
if you will have large amount of incoming mails. Maybe i am wrong ;)
> OK so if I have a global database but separate IMAP spame boxes how should i
> go about actually running sa-learn.
from the cron point of view I would have to enum all the users and do
/home/user/isSpam, /home/userb/isSpam, etc.
I could also run it in procmail, this would run more often but on smaller
files. What are my other options, which
is best.
>
>
> Alexey-12 wrote:
>> I prefer single(global) DB because all users get the same spam, and I learn
>> from any spam and "secure" with this
>> others - IMO it is a good idea.
>>> What is the best way to run sa-learn with separate IMAP training folders per
>>> user, but global bayes database.
>> --
>> Best regards, Alexey.
--
Best regards, Alexey.
