Re: SMTP 521 Reply code

[Harald.T.Alvestrand@xxxxxxxxxx]

At the MAILEXT working group meeting at the last IETF, the draft
got a rather unclear treatment.
The questions raised were:

- Is it enough? Some clients (in violation of spec) ignore the status
  code of the banner.

- Is it the Right Thing? It seems that a TCP-level function would be
  better, but TCP:ng is a long way off (if ever).

One simpler (?) solution is a server which replies "521" or another
500 code to *every* command; this would not need a spec change.

A further philosophical issue: Some PC mail gateway vendors might in
fact be tempted to try to use the existence of this extremely simple
"SMTP server" as a lever to wiggle out of the requirement that
"postmaster" should work on all servers. This is evil.

Well - anyway, the right list is mailext@cs.wisc.edu.

                 Harald T. Alvestrand