[...] I think what he is actually saying is that if you have a
machine that says "EHLO some.name" then there should be both a MX
record for some.name and a SRV record for _client._smtp.some.name
(i.e. a CSV/CSA record).
However, the standard requires that it says "EHLO host-at.some.name".
Not quite. It requires that the HELO/EHLO argument be a valid name for
the SMTP client host. The presence or absence of any DNS zone cuts in
the vicinity is completely irrelevant.
It is a seemingly simple task to drop the leftmost label(s) so as to
obtain the mail domain, but doing that properly requires a zone cut
algorithm that most servers miss.
...and which is wrong anyway. The division of DNS names into "hosts"
and "domains" is purely a human one. Dropping the first label from a
DNS name in an attempt to get "the domain" for it is, at best, a rough
heuristic. Looking up the DNS tree for zone cuts also is nothing more
than a heuristic.
It's not even clear to me that there *is* a "_the_ domain". What's
"the domain" for (to invent an example) mail.research.tjw.ibm.com?
There plausibly could be as many zone cuts as there are dots, there,
and I could argue for picking any of them as "the domain" for email
responsibility purposes (well, possibly excepting the TLD, but even
that is just a heuristic, likely to break soon).
/~\ The ASCII Mouse
\ / Ribbon Campaign
X Against HTML mouse(_at_)rodents-montreal(_dot_)org
/ \ Email! 7D C8 61 52 5D E7 2D 39 4E F1 31 3E E8 B3 27 4B
_______________________________________________
Asrg mailing list
Asrg(_at_)irtf(_dot_)org
http://www.irtf.org/mailman/listinfo/asrg