On Wed, 2010-10-06 at 00:35 -0400, Alex wrote:
> > > We _really_ need to change that rule's description...
> >
> > Uhm, while I would never argue that naming to be unfortunate in
> > hindsight, despite most of the time actually matching its stated goal...
> >
> > I blame this one on Alex (the otherwise anonymous $mysqlstudent). He's
> > been around long enough, by far, to know about this. Just simply and
> > occasionally glimpsing threads on this list should have told him.
>
> Yes, my fault. I have experimented with it in the past on smaller
> systems, but never wanted to implement it on any system that was
> particularly critical based on what I've read here.
>
> I had recently implemented it on a smaller production system based on
> some documentation that I read outside of spamassassin.org, that also
> talked about using bayes with mysql.
>
> Hope that helps to explain what happened, and I'll be sure to read
> more thoroughly before implementing on my larger production systems.
Well, AWL is after all a rather dumb per-sender per-net-block score
averaging system. If it is suitable for you, depends.
Yes, it *can* average down an occasional spammy message, sent by an
otherwise known to be good sender. Its stated goal.
However, even if most spam does not re-use senders *and* net-blocks, it
still can average up (or down, the bad thing one notices eventually)
spam. Just as it can average up ham -- rarely noticed, unless it crosses
the threshold, though.
Again, a score averager. Likely to never have a bad influence on ham,
unless someone sent a GTUBE previously. Helpful, to counter the
occasional spammy message by a good sender.
Anything else is just the threshold crossing F[PN] you otherwise
wouldn't even have realized AWL fired on.
--
char *t="\10pse\0r\0dtu...@ghno\x4e\xc8\x79\xf4\xab\x51\x8a\x10\xf4\xf4\xc4";
main(){ char h,m=h=*t++,*x=t+2*h,c,i,l=*x,s=0; for (i=0;i<l;i++){ i%8? c<<=1:
(c=*++x); c&128 && (s+=h); if (!(h>>=1)||!t[s+h]){ putchar(t[s]);h=m;s=0; }}}
