Dave Crocker wrote:
On 5/1/2014 2:51 PM, Mark Andrews wrote:
In message <5362B4C6.10904@xxxxxxxxxxxxxxxx>, Miles Fidelman writes:
Dave Crocker wrote:
5. The IESG has concluded that this document extends an IETF protocol
in a way that requires IETF review and should therefore not be
published without IETF review and IESG approval.
Since DMARC does not extend any existing IETF protocol, how is that
reference useful here?
Sure looks to me like DMARC extends both SMTP and DNS.
And DKIM.
No, No, and No.
Don't confuse 'use' with 'extend'.
TCP "uses" IP. It does not "extend" it. That's what architectural
layering is about. Functions above a layer do not extend the lower
layer; they use it.
DMARC /uses/ DKIM (and SPF). It does not alter (extend) either of them.
As for any claim that DMARC 'extends' SMTP or DNS, it's difficult to
imagine the technical logic behind such an assertion.
1. DMARC essentially species a change to the behavior of mail
reception. It sure looks to me like an extension to the state machine
model for mail processing, at the SMTP level. Arguably, the same can be
said for what forwarders (including mailing list processors) do.
2. As to DNS: There's a long standing argument about the use of TXT
records for purposes beyond holding "descriptive data." Last time I
looked (just now) there has been nothing since RFC1035, which states
"TXT RRs are used to hold descriptive text" and RF1464, about storing
arbitrary string attributes, and remains experimental. DKIM, SPF, and
DMARC both define and use TXT RRs in ways that go beyond their defined
scope.
d/
ps. The original note was from Jim Fenton and it was him I was asking
to explain his reference. He seemed to be making a point and I was
asking him to provide it explicitly.
Yes. I'm agreeing with Jim, and providing some backup to his point.
Miles Fidelman
--
In theory, there is no difference between theory and practice.
In practice, there is. .... Yogi Berra