Re: [Ntp] I-D Action: draft-ietf-ntp-mode-6-cmds-09.txt

[Daniel Franke <[e-mail]>]

On Tue, Jun 23, 2020 at 1:17 PM Salz, Rich
<rsalz=[e-mail]> wrote:
>
> The IETF does not have the concept of a reference implementation.  Maybe it used to.  But for a WG to use the term "reference implementation" it implies a cachet, at least, that is at odds with some of the core principles of interop and RFC's.  The IETF does not distribute source code, therefore it would also be wrong for the IETF to name someone as a reference implementation.

RFC 5905, Appendix A, does contain some code. I think it is
appropriate to refer to this code, and *exactly* this code, as a
reference implementation. It is also appropriate to acknowledge, as a
simple undisputed fact, that this code was historically derived from
the NTF's implementation. It is inappropriate to refer to any *other*
code as a reference implementation — not even other NTF code from the
same release that the code in Appendix A came from — regardless of how
well it may or may not adhere to the text of RFC 5905.

As a broader philosophical note, RFC 5905 is not the only example of
IETF including code in an RFC, but it's something we do sparingly, and
for good reason. Specifying a protocol only through an implementation
leaves it unclear what it is actually meant to be required for
compliance with the spec and what is an incidental implementation
detail. It should be done only as a last resort when the authors have
tried and failed to clearly communicate their intent in other ways.