Re: Issue: Transition rules?

[James M Galvin <galvin+ietf-nomcom@eListX.com>]

On Wed, 15 Oct 2003, Robert Elz wrote:

      | With protocols I believe we all expect that when a revision is
      | published it will be adopted immediately and implementations
      | will transition as soon as possible.

    Huh?  You mean that if I have an existing SMTP connection, and a new
    doc is approved that alters the SMTP dialog, you expect my currently
    running processes to adapt to the new procedure and enforce the new
    rules?

Of course not.

The point is, when a revision is approved it is presumed to be the
authority from that point forward.  Does that mean we sit in judgement
of those who don't *immediately* upgrade?  Of course not.

When transition is an issue we account for the explicitly in protocols.
We either ensure backward compatibility (to the greatest extent
possible) in the new protocol or we accept that there will be two valid
protocols for a period of time and we deal with it in some other way.

The principle is still that the new document is the authority.

Why shouldn't that be the case here with the additional caveat that we
don't revisit earlier decisions?  In essense, not revisiting prior
decisions (past events) is the transition plan or backward
compatibility.

    If what you're suggesting is that an existing nomcom should be
    (metaphorically) killed, and then reconstituted under the new
    procedures, then fine, that would be a possibility - but I don't
    think it is one that anyone would consider very seriously.

Again, of course not.  I'm suggesting the NOMCOM would simply continue
as it was but it would adopt the new document rules and procedures from
the point that it was approved.

Jim