On Mar 16, 2021, at 1:26 PM, Dave Crocker <dhc(_at_)dcrocker(_dot_)net> wrote:
Recent list discussion prompted me to look back for earlier reference to the
first digit 'rule', concerning test in Section 4.2 of RFC 5321.
The above logic escapes me.
While it gives direction when the full code is not (yet) understood, it says
nothing about adoption processes that might or might not eventually lead to
support for new codes.
That is, new codes will of course not be understood by any implementation
initially, but useful codes are likely to be understood by 'most'
As such, the first digit rule, in RFC 5321, really has nothing at all to do
with the efficacy of adding code or adding text to codes. It only has to do
with behavior of implementations that do not (yet) understand a full code.
Postfix has been around since 1997 (alpha), or if you prefer 2001 (1.0 prod).
The SMTP client in Postfix only ever looks at the first digit. I expect this
is fairly common. There isn't yet any known response code where we'd expect
the rest to matter. I'd prefer to keep it that way. The remaining digits are
useful for forensics.
 There is one special case, 421 in response to EHLO is taken as a no
service indication, rather than a reason to try HELO instead. This
should probably include 521 (perhaps an oversight).
ietf-smtp mailing list