Eric A. Hall writes:
On 6/16/2004 5:36 AM, Arnt Gulbrandsen wrote:
I'm pretty sure we'd be doing a SRV lookup for _marid._udp.<domain>
and then asking that server whether to fail/pass/... the message.
Has this been proposed? I'd support it since it avoids most of the
problems with the existing approaches.
I've mentioned it once or twice. Noone else has, as far as I know.
Wasn't quite sure whether I was overlooking something obvious or everyone.
It could still be made extensible if the RR listed the servers and the
needed test data (MAIL-FROM, From:, Message-ID:, Date:, ...). The
size of that RR would not be prohibitively large.
That would be possible even without a new RR, if the server's allowed to
return a result such as "cannot decide for lack of header fields x y
and z". But if the server's allowed to ask for any amount of data, TCP
has to be used.
Arnt