On 29 May 2003, Nigel Metheringham wrote:
> > And I was wondering if it would be possible to have exim behave like this
> > RFC states?
>
> I'd say its a little early yet... there are 2 alternative DNS syntax
> proposals there. I would prefer to have the RFC at least in a later
> draft form before releasing any implementations.
>
> Could anyone enlighten me how much needs doing to resolvers and DNS
> servers to support this? I assume there will need to be some degree of
> underlying infrastructure rehashing to allow this stuff to be produced.
> [Unless you hack it into the TXT records]
>
> However it looks to me as though it would be relatively easy to support
> within exim acls. In the early stages though it might be better done as
> a scoring mechanism in spamassassin - ie add 4 points for incorrect RMX
>
> We do need to work out appropriate exim acl syntax.
ACL+SpamAssassin only cover half of the proposal.
Section 6.3 requires exim to change the envelope sender address
on mailing list and .forward'ed mail (so that the mail comes from
a domain "RMX served" by this machine).
I wouldn't be suprised if this can be done without changing exim,
perhaps with header rewrites or headers_rewrite, but it may be better
to add specific support inside exim, if only to reduce the likelyhood of
mis-configuration.
The current spec wishes to disallow empty sender address,
but merely states that an alternative solution must be developed for
bounces and other anonymous mail.
I'd like to see a concrete solution for bouces before using
this proposal in anger.
--
Dr. Andrew C. Aitchison Computer Officer, DPMMS, Cambridge
A.C.Aitchison@??? http://www.dpmms.cam.ac.uk/~werdna
