Stories
Slash Boxes
Comments
NOTE: use Perl; is on undef hiatus. You can read content, but you can't post it. More info will be forthcoming forthcomingly.

All the Perl that's Practical to Extract and Report

The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
 Full
 Abbreviated
 Hidden
More | Login | Reply
Loading... please wait.
  • I'm sure you've considered the idea already, but I'm curious why you don't want to use spamassassin. My server is running spamassassin version 2.60 and but oh my goodness, this version kicks butt. I typically get 2-3 spam a day that are false-negatives; versus 350 a day that are caught. (heh, checked that; it's up to 385 a day now). I occasionally check for false-positives and there have only been two in the last two weeks; both caused when I falsely tagged a message as spam in the bayesean filter.

    The only user-level customizations I've done are: 1) auto-predjudice all mail containing microsoft executables as spam; 2) told it I only read english language/locales; and 3) whitelisted two addresses (but one is my mom).

    My only behaviour I've needed to change is: saving false-negative spam into a folder so I can regularly feed it to 'sa-learn' to improve the bayes network.

    Setup on the server was a piece of cake. The only real choice I had to make was whether to run it at user-level, or as a system daemon. I'm running the daemonized version, directly hooked into my MTA (exim; though it works just as well with sendmail and qmail). I didn't have to install any separate RBL-checking (which is built in) or tools like Virpul's Razor (which is an option, but I decided took too long to process each message). It takes 40 seconds realtime to pass an email through, but adds almost no server load (with 10 active users on the server). I think it spends most of its time waiting for results from the RBLs.

    --

    -DA [coder.com]