On 19 nov 2006, at 04.37, Cullen Jennings wrote:
On Nov 14, 2006, at 11:03 AM, Paul Hoffman wrote:
At 4:17 PM +0100 11/14/06, Joe Abley wrote:
For the benefit of those who do not follow dnsext closely, what
friction do you expect?
As Eric stated in his message, we should not rehash old arguments.
This has been beaten to death on the DKIM WG mailing list. As
expected, different people had different (and, in this case,
strongly-held) views, but consensus was reached and agreed to by
the AD and with the DNS folks.
To avoid repeating this debate, can someone post some summary
information on this particularly including which exact people came
to consensus about this. I'm particularly interested in if the
consensus included the contributors to draft-iab-dns-choices since
that has been raised in LC comments.
The draft-iab-dns-choices document, that I am one of the editors of,
is waiting for the draft-ietf-dnsext-2929bis document that specifies
a more clear process (and easier) for new RR types.
I have personally told everyone that ask, including the DKIM people,
that the best solution is to create a new triple {owner, class, type}
as specified in the choices document, so the RRSet returned as
response to a DNS query is containing the records one want, and as
small number of non interesting records as possible.
My personal, most personal, view is that creation of a new RR, and
use of a new RR, is easier than many people think.
The creation of a subdomain in the case of DKIM do make a new triple,
and make the deployment easier, as the subdomain that the DKIM
related data is stored in can be delegated and managed by the mail
administrators of the main domain, while the main zone is managed by
the DNS people. So, I think the creation of the subdomain is a good
choice regardless of whether a new RR type is created or not. At the
same time, that subdomain most certainly will contain non DKIM
related data as well, for example data related to DNSSEC.
But, as you ask:
I was part of the discussion in the DKIM wg, and I do accept the
rough consensus was to move forward with reuse of an existing RR Type
(TXT).
Note that the draft say:
3.6.2.2 Resource Record Types for Key Storage
The DNS Resource Record type used is specified by an option to the
query-type ("q=") tag. The only option defined in this base
specification is "txt", indicating the use of a TXT Resource Record
(RR). A later extension of this standard may define another RR
type.
I do not see any need for reopening this question again. DKIM need
this document, NOW!
But, I personally do hope someone will create a new RR Type
definition for DKIM some day in the future.
Patrik
_______________________________________________
Ietf@xxxxxxxx
https://www1.ietf.org/mailman/listinfo/ietf