[Mimedefang] SpamAssassin 2.40 experiences?

[Nels Lindquist]

On 4 Sep 2002 at 12:40, David F. Skoll wrote:

> I downloaded SA 2.40, and tested it against a "canonical" spam
> message I keep for that purpose.  Against 2.31, the message scored
> 9.4, but against 2.40, it scored -0.10!  Needless to say, I was
> shocked!
> 
> Do people find the weights in 2.40 are accurate, or is there a problem
> here?  I'm sticking with 2.31 for now...

Haven't been able to test yet, but there's currently a discussion on 
the SA list which might be germaine.

How did you test 2.40? 

If you forwarded the message to yourself for processing by MD/SA 
(which is what I generally used to do) your MUA may have added 
Resent* and References* headers, which are newly recognized by SA 2.4 
as being strongly indicative of non-spam.

IOW, if you forwarded spam to yourself, SA may have correctly 
determined that it was sent by a non-spammer. :-)

----
Nels Lindquist <*>
Information Systems Manager
Morningstar Air Express Inc.