Thanks Dan. Actually, after more investigation (I spend too much time investigating) I
found milter-spamc, which does what I want with the -b and -c flags. I can redirect
spam
and non-spam to separate mailboxes.
I can then use these for learning if I wish, once I'm satisfied they contain spam or
non-spam as appropriate.
Thanks,
Matt
Dan Nelson
<[EMAIL PROTECTED] To: Matt Tencati <[EMAIL PROTECTED]>
group.com> cc: [EMAIL PROTECTED]
Subject: Re: [SAtalk] Re: Can base64
encoded messages learnable by
09/04/2003 sa-learn?
10:00 PM
In the last episode (Sep 04), Matt Tencati said:
> I've been looking for a milter to do the caching of messages - do you
> know of one?
So far our spam load hasn't been bad enough to require bayes, but the
sample milter in the libmilter documentation could easily be modified
to log into a database instead of textfiles. You could even use it the
way it is. Just modify it to rename the tempfile to something like
msg.timestamp.senderaddress for easy lookups, and age the directory
with a "find . -mtime +2d -delete" every night.
http://www.they.com/doc/libmilter-dev/libmilter/html/sample.html
Now it looks so easy I'm going to have to try it myself :)
--
Dan Nelson
[EMAIL PROTECTED]
-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_______________________________________________
Spamassassin-talk mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
