[SAtalk] Re: Subject contains username

2004-01-02 Thread Mark London
Theo Van Dinter wrote: On Thu, Jan 01, 2004 at 09:33:23PM -0500, Mark London wrote: >Hi - I see the rule USERNAME_IN_SUBJECT, but I've never seen that triggered at my site, and a test message sent to myself did not trigger it. We are using spamassassin with sendmail, mimedefang, cy

[SAtalk] Re: Subject contains username

2004-01-01 Thread Mark London
04 21:27:44 -0500 To: [EMAIL PROTECTED] From: Mark London <[EMAIL PROTECTED]> Subject: london, you look cute. Content-Type: text/plain; charset="us-ascii" X-Spam-Level: 0 () X-Scanned-By: MIMEDefang 2.31 (www . roaringpenguin . com / mimedefang) Theo Van Dinter wrote: On Wed, Dec 31,

[SAtalk] Spamassassin rules suggestion, where to talk about?

2003-12-30 Thread Mark London
Are suggestions and discussion regarding spamassassin rules discussed here? The GUARANTEED_100_PERCENT tested should be expanded to phrases like "100% safe" "100% satisfaction", etc. Some spam about HGH are getting through, because the HGH are being spaced out with several non-ascii characters.

[SAtalk] Mimedefang/spamassassin/base64 encoded text

2003-10-06 Thread Mark London
I'm sorry if this questioned has been already answered here, but I couldn't find it using a google search. I've been using the following configuration for a couple of months: Mimedefang 2.31 & Spamassassin 2.55 I just received a spam message that has base64 encoded text, and the spamassassi

[SAtalk] Suggestion for new rule for spam getting through.

2003-10-02 Thread mark london
I got the following spam message, which has a low rating. The construction of the To: line is "To: user <[EMAIL PROTECTED]>"I've seen this in a lot of spam, and can't imagine how a real message sent to me would have this form, so I would like to create a test for this situation. Has anyon

[SAtalk] Spamassassin algorithm question.

2003-09-20 Thread Mark London
Does spamassassin simply total up the scores of all the tests which are found to be valid for a message, or does it increase the score based on whether a combination of tests exist? I.e. the likelihood that a message is spam if it only tests positive for the GUARANTEE test might only be someth