From: "Matt Kettler" <[EMAIL PROTECTED]>
At 11:11 AM 10/13/2005, Andy Hester wrote:
Thanks for the info.
Since I really just want to block email with certain words is there a
reason I shouldn't just put some rules in header_checks and body_checks in
postfix?
Would this be better or worse?
If you want to do an unconditional block of all email containing certain
words, doing it in your MTA is definitely better. Of course, unconditional
blocks should be treated with caution. "anal" is a substring of "analog",
so set your blocks up VERY carefully lest you wind up blocking emails
talking about unrelated subjects.
SA is for when you want to tally up scores and only block emails that
reach a threshold. Yes, you can "force" it to block mail by setting the
score of a rule really high, but it's generally more efficient and easier
to set this up at the mta layer.
This is not necessarily a good thing to do. "Breast" and "teat" are
bad words to some people. But they can get in the way of discussions
of a form of cancer that some men get and a lot of women get or of
milking cows.
Even the dreaded V word drug has practical uses in treating some non-
<E-word> medical problems that are life threatening.
It is best if the scores are added up to form a cluster of bad words,
B-word, P-word, F-word, and so forth. Also you are in best shape if the
obscured forms of the words are scored heavier than the clean text
versions.
And be sure to remember, that one person's religious problem with a
given word is another person's serious life threatening medical problem.
I don't have the arrogance to function as an absolute censor based on
single words found in text.
Now, I'd be DELIGHTED if each rule had a number associated with it for
the number of times the rule was hit. That MIGHT be useful for some
forms of this level of spam filtering.
Now let's see if Apache's stupid filters lets this through.
{^_^}
