Re: Nomination for promotion to "Proposed Standard"

[ned.freed@mrochek.com]

> The gating factor the IETF controls is actually DNS RR numbers - and that
> is in DNSEXT.

This approach might work for the document in hand, but the notion that this
gives us general control is an illusion. Consider SPF. As currently written it
uses TXT records. There have been proposals to use a new RR type with SPF, or
to reuse MF or MR, but the reception for these has been lukewarm at best. If we
attempt control through RR allocation all we'll do is insure that this deploys
as some form of TXT record. (Which may or may not be a bad thing. My point
isn't to advocate a particular approach, but rather to caution against the
notion that these schemes are dependent on new RRs being deployed.)

> I talked to Ted and Steve (among others) about the RMX thing in Chicago,
> and agree fully with your analysis of "we need to do something", with
> "document the cases for which it is broken" being high on the list of
> "somethings".

Harald, as I said before, I have serious misgivings about any approach that
puts us in the position of opposing such schemes unconditionally. Any document
we produce in this space needs to examine both the features and the faults of
such schemes. Focusing on the potential damage they do would be a mistake IMO.

More generally, I think the question we all need to keep in the back of our
minds is the very real possibility of widespread deployment of even more
draconian schemes. For example, what if all the major service providers were to
enact schemes similar to the one AT&T used recently?

> (idle query: does @acm.org mail forwarding change envelope FROM?)

Likely not, but amusingly enough the forwarding Exchange typically does
overrides the envelope from, and has a tendency to create loops in the
process. This can be seen as a predictor of second order problems with
some sorts of envelope from override schemes.

For my own part I long since recognized the possibility of one of these schemes
deploying, and I've taken steps in the software I write to provide a variety of
envelope from override schemes that attempt to work around both the first and
second order problems in this space.

> But I can't fight all battles - so I flipped this one in the direction of
> the AD who is in charge of the WG that is attached to the control point.

I guess I'm OK withi this as long as we don't attempt to us this appproach to
control either RMX or SPF.

> I'd like Thomas to say "no, and here's why" - or get someone else to do
> that. (Ned, I didn't want to ask you to take charge of this rampaging
> beast, given that you won't be in Minneapolis, and have other serious
> distractions....)

Indeed.

				Ned