My hold weight is 15, I don't worry about the negetive weights right now.
Once my cpu optimization is completed and URI scanning is done, I am going
to change my focus on improving spamassassin.
I can tell you this, from stats, that the use of spammassassin with my other
tests has reduced my spam by 33%.
spamc32 normally adds 1 to 6 points to spam messages, thereby, allowing me
to reduce the scores on my other tests, reducing false positives, and
further punishing spam. Since I have done this, I have received emails from
mailing lists that I used to have signed up for years ago :)
Contact me off list if you want further details of my config file.
I love this declude system, and spamc32, it has become quite the hobby to me
:)
Travis
----- Original Message -----
From: "Geoff Varney" <[EMAIL PROTECTED]>
To: <[email protected]>
Sent: Monday, October 24, 2005 1:21 PM
Subject: RE: [Declude.JunkMail] SPAMC32 -e option in JunkMail
Thanks Travis. I have now begun to see some results in my declude log
file
that shows what I did is passing the score on to declude. However, the
problem there is that I have set the SPAMASSASSIN action to add a footer
that the message is likely SPAM, and it adds that when the test is
"failed",
even negative scores from SA.
What is your hold weight? I have my routeto (a SPAM folder I review)
weight
at 10 which doesn't give to many false positives. I wonder what your hold
weight is so that I can tailor your command line to suit my "hold" weight
if
different than yours. Is there a way to ensure that the SPAMC32 weight
passed to JM doesn't go negative?
Geoff
-----Original Message-----
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Travis Sullivan
Sent: Monday, October 24, 2005 11:01 AM
To: [email protected]
Subject: Re: [Declude.JunkMail] SPAMC32 -e option in JunkMail
Hi Geoff, this is what I use:
SPAMASSASSIN external weight "e:\imail\declude\spamd\spamc32.exe -d
#serverIP# -cw %WEIGHT% -sw 15 -e -et 11 -f"
I don't want to check spamassassin if my hold weight is already reached,
and
I don't want the message to weigh more than 11 from spamassassin, prevents
one test from failing a message, thereby reducing false positives.
This allows the score to come from spamassassin, not assigning 10 points
to
all failed spamassassin tests. Most of spam assassin tests only return 1
to
5 points. It is working for me quite well.
Travis
----- Original Message -----
From: "Geoff Varney" <[EMAIL PROTECTED]>
To: <[email protected]>
Sent: Monday, October 24, 2005 12:53 PM
Subject: [Declude.JunkMail] SPAMC32 -e option in JunkMail
Hi,
I am finally getting SpamAssassin working for my JunkMail with SPAMC32
(0.5.57) calling SPAMD from a Linux box. With training of the Bayesian
databases well underway I am seeing significantly more SPAM being caught.
However, I'd like to verify the way the SPAMASSASSIN entry in the Declude
global.cfg works. Here is what I had originally:
SPAMASSASSIN external nonzero
"c:\imail\declude\filters\spamc32\spamc32.exe -u spamd -d <IP of SPAMD
server> -cw %WEIGHT% -sw 10 -y -f" 10 0
This works fine, but the weight of 10 then assumes all SpamAssassin
identified messages are SPAM since Declude considers 10 the score for
SPAM.
So I thought I'd use the -e option for SPAMC32 to pass the actual SA
score
to Declude to possibly minimize false positives.
Here is the line I have now:
SPAMASSASSIN external nonzero
"c:\imail\declude\filters\spamc32\spamc32.exe -u spamd -d <IP of SPAMD
server> -e -cw %WEIGHT% -sw 10 -y -f"
Is this correct usage? By using -e and dropping the weights at the end
of
the command line does this then pass the SA weight for Declude to apply
to
the total weight? I think this is what's happening now, but wanted to
check
with someone who has used this for a while and knows it better than I do.
Thanks,
Geoff
---
This E-mail came from the Declude.JunkMail mailing list. To
unsubscribe, just send an E-mail to [EMAIL PROTECTED], and
type "unsubscribe Declude.JunkMail". The archives can be found
at http://www.mail-archive.com.
---
This E-mail came from the Declude.JunkMail mailing list. To
unsubscribe, just send an E-mail to [EMAIL PROTECTED], and
type "unsubscribe Declude.JunkMail". The archives can be found
at http://www.mail-archive.com.
---
This E-mail came from the Declude.JunkMail mailing list. To
unsubscribe, just send an E-mail to [EMAIL PROTECTED], and
type "unsubscribe Declude.JunkMail". The archives can be found
at http://www.mail-archive.com.
---
This E-mail came from the Declude.JunkMail mailing list. To
unsubscribe, just send an E-mail to [EMAIL PROTECTED], and
type "unsubscribe Declude.JunkMail". The archives can be found
at http://www.mail-archive.com.
