Tony Hansen wrote:
(ii-a) Make it clear that the codes may be different and that
clients are expected to respect the *last* value listed
I'd feel better with "SHOULD NOT be different". But if they're
nevertheless different taking the last code for real as suggested
in RFC 2821 is the most obvious solution for this odd situation.
P.S. to Hector: An SMTP extension specified in an RFC, offered
by the server, and explicitly accepted by the client, is a good
excuse to violate this general SHOULD NOT addressing backwards
compatibility. IOW you can use different codes if both sides
agree - ideally you'd specify it in an RFC. You can't claim that
2821bis allows this _without_ mutual agreement, it would violate
the proposed SHOULD NOT.