At 07:56 PM 11/22/03 -0800, John Oliver wrote:
> 1) The Swen virus is not spam. It should be caught by anti-virus
> software, not by anti-spam software.

It's email that I don't want.  It has a predictable From: and/or
Subject: and so should be able to be scored very high.

Agreed that swen emails may be undesirable, but as others on the list have pointed out, that alone does not make them spam.


SpamAssassin targets spam as loosely defined to mean unsolicited commercial email and scam email. By very strict and long-standing policy SA is developed with a neutral attitude towards viruses. If they get tagged, that's fine, if not, no big deal either.

Quite frankly, there's no need for SA to target viruses, and because of how SA works it cannot be updated rapidly to keep abreast of the latest virus releases.. This IS the domain of virus scanners.. it's what virus scanners are for, and there are lots of them, including free ones like ClamAV.

SA's development resources are already stretched a bit thin, so it's quite futile and damaging to the project to dilute the development effort expanding SA to do things that other software does VERY effectively. Even diverting all of SA's development resources to virus catching would result in something less effective than even the worst virus scanner. On the other hand it would also make SA a horrible spam scanner.

Hence the policy.. let the virus scanner writers deal with viruses, because they are already doing a good job of it, and let SA focus itself entirely on spam. And this ultimately leads to better products.. unlike the common "swiss-army knife" approach to software (ie: microsoft word which tries to be everything), many open source projects focus themselves on doing one thing VERY well, instead of trying to do lots of things in a half-assed manner.

Which would you rather have, a single software package that's a lousy virus scanner and a mediocre spam scanner, or a damn good spam scanner? I vote for the latter, since I can always add on clamav if I need virus scanning.

Now, none of this is to say that SA can't handle these emails.. it's just not any part of the "out of the box" focus of the tool.

Apparently not.  It looks to me like something has to be done other than
install SA to make them go, and I'm working on figuring that out now...
:-)

Well, SA works so-so out of the box.. but there's definitely some trade offs that you can do to make it more effective.


DNS blacklist checks help quite a bit, but the drawback is added processing time for the DNS lookups.. Install the Net::DNS perl package and the DNS blacklists should start being checked automatically.

Vipuls razor is a nice add on, and pyzor and dcc are similar tools. All you need to do is install them and SA should start using them transparently.. these tools are also network-check based, so they have the same tradeoffs as the DNS blacklists.


bayes is a great help, but requires some care and feeding, and can consume a moderate amount of disk space depending on how much mail you run. This is also one area where you can fairly easily teach SA about what YOU consider spam, as opposed to what it is pre-configured for.


Don't like swen worm mails? Train them as spam with sa-learn

Have some annoying false positives? Train them as ham with sa-learn.

Since bayes is trained to YOUR email, as opposed to anyone else's, it can really wind up with a behavior that adds a lot of "custom fit" to SA's behavior.









-------------------------------------------------------
This SF.net email is sponsored by: SF.net Giveback Program.
Does SourceForge.net help you be more productive?  Does it
help you create better code?  SHARE THE LOVE, and help us help
YOU!  Click Here: http://sourceforge.net/donate/
_______________________________________________
Spamassassin-talk mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/spamassassin-talk

Reply via email to