Andrew Newton wrote:
On Apr 22, 2004, at 8:49 AM, Yakov Shafranovich wrote:
I think that it is pretty clear (to me at least). When a sender
publishes MARID records, there are two specific things the sender can
choose from: either 2821 values or 2822 values (or both). The receiver
is provided with two specific validation paths: one for 2821 and
another for 2822. So it is not a free for all but rather we narrow
things down to two very specific paths, both on the sender and
receiver's parts.
Just a clarification, in case I misstated something (or don't
understand something). I proposed that the sender does not specify the
type of identity (2821 vs. 2822) in the qname or rdata. I think Greg
did a pretty good job explaining the reasons for it. But to add to
that: if the sender can choose 2821 and the receiver uses 2822, then the
two have failed to communicate.
I assumed that what you are saying is the following:
1. Sender publishes MARID record.
2. Sender can either use 2821, 2822 or both.
3. Receiver will check 2821, if that does not verify then move on to 2822.
Yakov
-------
Yakov Shafranovich / asrg <at> shaftek.org
SolidMatrix Technologies, Inc. / research <at> solidmatrix.com
"And this too shall come to pass"
-------