On Mon, 26 Jan 2004, Paul Diaguila wrote:
> No Bayes db yet, but I would think the one rule would score it a 5
>
> Paul
>
> Covington, Chris wrote:
>
> >Your Bayes must be hosed if what you think is spam gets BAYES_00.
> >
> >Chris
[snip..]
> >Greetings
> >
> >Using SA Ver. 2.63 with Mimedefan
At Mon Jan 26 16:12:30 2004, AltGrendel wrote:
>
> > describe SUBJECT_ENCODED_MY_TEST Subject begins with =?
> > scoreSUBJECT_ENCODED_MY_TEST 5.0
> >
> > When a message comes in:
> >
> > Subject:
> > =?ISO-8859-1?b?V2UgaGF2ZSB3aGF0IHlvdSBuZWVkIC0gQ2hlYXBlc3QgcHJlc2NyaXB0aW8vbnMgb24gdGhlIGl
On Mon, Jan 26, 2004 at 11:12:30AM -0500, AltGrendel is rumored to have said:
> >
> > scoreSUBJECT_ENCODED_MY_TEST 5.0
> >
> > X-Spam-Score: 3.422
> > BAYES_00
I think that's what's getting your score down.
> > ???
>
> Not sure what you're getting at here, try:
I think he d
Are you learning spam as ham? Your bayes_00 would
counter to give a negative score and that's why these messages are slipping
in. If you increase the score on your local test it will most likely catch
these but you most likely have a bad bayes db.
You might want to turn Auto-learning off
Your Bayes must be hosed if what you think is spam gets BAYES_00.
Chris
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Paul
Diaguila
Sent: Monday, January 26, 2004 10:44 AM
To: [EMAIL PROTECTED]
Subject: [SAtalk] too much spam...
Greetings
Us
No Bayes db yet, but I would think the one rule would score it a 5
Paul
Covington, Chris wrote:
Your Bayes must be hosed if what you think is spam gets BAYES_00.
Chris
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Paul
Diaguila
Sent: Monda
On Mon, 2004-01-26 at 10:43, Paul Diaguila wrote:
> Greetings
>
> Using SA Ver. 2.63 with Mimedefang, and still quite a bit of spam is
> getting through. Have all the current BigEvil, ect... As an
> example, a rule is in place in local.cf
>
> header SUBJECT_ENCODED_MY_TEST Subject:raw =~ /
