On Mon, May 5, 2014 at 1:17 PM, Dave Cridland <dave@xxxxxxxxxxxx> wrote:
The SMTP state machine is not changed by DMARC any more than SPF or others changed it. It doesn't add any new states, verbs, parameters, or anything else. DMARC sits at least two "layers" above where SMTP operates. As with any number of other filtering systems, it can influence SMTP's final DATA reply, but that's hardly unique or even unusual.
It aims to change the behaviour of Internet Mail as deployed.
So did sendmail rewrite rules, SPF, spamassassin, or anything else you can recall that established new acceptance requirements of some kind. This is hardly a new concept.
Whether you want to claim that this is formally extending SMTP, per-se, or not is really something of a moot point - there is certainly an intentional, large, effect on the deployed protocol. Arguing whether this fits the letter of some particular definition smacks of lawyering to my mind.
I never said that it has no impact, or even a small impact. (I would also point out that amid all this angry mudslinging, it's been pointed out several times that DMARC has been in use by some operators for more than a year, and for them it works fine and has had no visible negative impact.)
What I am saying is that it does not achieve its impact using the mechanism claimed up-thread, namely some kind of extension to SMTP or DNS.
What I am saying is that it does not achieve its impact using the mechanism claimed up-thread, namely some kind of extension to SMTP or DNS.
Speaking derisively about precision of _expression_ in a standards community seems pretty strange to me.
-MSK