On Wed, May 26, 2004 at 08:22:21AM -0700, Scott Call wrote:
> > From that, I infer that you would invoke "dspam" as a separate process,
> > and based on the results printed on stdout, classify the message as ham
> > or spam. That you can do using a "${run ...}" expansion, perhaps:
>
> Te problem with invoking ${run.. is that dspam uses per-message signatures
> (they look like "!DSPAM!5324hj532l32!" on the end of each message, which
> is what enables training.
> I think (I could be wrong) that the example you posted just scans the
> returned message for certain text and then denies it if the text exists.
If I understand it, Dspam can add headers for "Is Spam" starting with
v2.6.5:
X-DSPAM-Result: (Spam|Innocent)
X-DSPAM-Probability: (Floating Point Probability)
So you can later filter on these.
--
Avleen Vig
Systems Administrator
Doing virtual domain hosting with Exim?
Check out Virtual Exim: http://silverwraith.com/vexim
(Click the banner, support development..)
