> Some messages don't have message IDs... you might like to assume that
> this is an indicator of badness, but you may in some circumstances
> (notably very simple devices sending email) you will be wrong :(
> You will sometimes reject good messages if you do this. You may view
> this as acceptable and everyone applauds because you block more spam, or
> you may have users who will suggest that you should be sacked from your
> job because the message they wanted to receive was rejected....
yes, i agree with all these points of view. It's not easy to find a good
compromise between RFCs and real MTAs :(
> But by all means let a Bayesian spam filter see the headers and include
> this as one of the factors that is taken into consideration.
>
So, what you suggest for Bayesian spam filter?
In addition of several checks provided by Exim, we are using greylisting,
SpamAssassin and Anti-virus filtering whit Clamav.
Indeed, we achieved a high rate of spam filtering, but i think there is
always better to do :)
Paulo
--
Escola Superior de Enfermagem do Porto
Rua Dr. António Bernardino de Almeida
4200-072 Porto - Portugal
Tel: +351 22 5073500 - Fax: +351 22 5096337
http://portal.esenf.pt
