better still would be to abandon your domain-name matching scheme in
favor of more general heuristics.
I wasn't clear about this before, but I do have a bunch of heuristic
recipes above the domain-matching section. Probably 90% of the matches
are grabbed by the first 30 lines of my .procmailrc. But the heuristics
aren't 100% reliable, either. The combination of both heuristics and
domain matching has reduced my spam level to almost nothing (5 a month,
roughly). This was *before* AGIS pulled the plug on their favourite
spamming customers, by the way. I've received one lonely spam since the
15th, and that was happily filtered by my
:0 h
* ^Comments:.*Authenticated sender is
* ! ^X-mailer:
procmail/spamfile
recipe.
Other people that use the "match domain" method do some kind of a grep
on a file; if you're going to take this route, it might be something
you want to consier.
Amazingly(?), running through the 1900+ recipes only takes about 1/4 of a
second using procmail. I'd think there'd be more overhead involved in
forking off a grep, at least while the number of domain matches are
"reasonable" (I never thought I'd hear myself say that 1900 spamming
domains would be reasonable).
Anyway, thanks to everyone for their suggestions on tightening up my
rules! Like I said I'm new to the procmail game, and there are some neat
tricks available that aren't obvious from the procmailrc/procmailex
manpages!
Speaking of learning, does anyone have the O'Reilly regexp book? Is it
worth buying? I can't find it locally, so will probably buy it over the
phone, soon.
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
| Jason Marshall, marshalj(_at_)spots(_dot_)ab(_dot_)ca(_dot_) Spots
InterConnect, Inc. Calgary, AB |
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-