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
Stories, comments, journals, and other submissions on use Perl; are Copyright 1998-2006, their respective owners.
Viruses (Score:2)
Heuristically detecting viruses is a bitch of a problem. Well not exactly a bitch, it's actually far easier than detecting spam because you can determine that something is trying to do something malicious (even in PE code) (I had a talk about this for OSCon, but it was rejected - we do all of this in perl). Whereas with spam we have to detect the malicious language, if you like. The problem with viruses is you have to decode stuff - PE files, Word documents, Javascript, etc. That's the hard part.
So the plan for SpamAssassin is to not do virus detection, ever. At least for now.
Reply to This
Re:Viruses (Score:1)
Still, can't complain--it works as advertised, and rather well at that. And thanks to Nat, I managed to abuse it into doing things it didn't advertise, so I'm happy. It's on my permanent list 'o cool system uti
Re:Viruses (Score:2)
Hey don't say that! If it wasn't for microsoft and their great security I wouldn't have a job
I'd give you the key to detecting all Klez variants if I knew exactly what it was myself, but I can't grok our AV guy's code. It's something to do with disassembling the PE code (Win32 binary) and detecting it based on the lack of the CompanyName header in there. Or something like that.
Also of interest may be OpenA [openantivirus.org]
Re:Viruses (Score:1)
I threw a rule with a weight of 5.6 in my SpamAssassin config files for messages with a Content-ID: header in the body. Rumor has it that it's not a 100% guarantee, but it's working as a filter trigger for me.