On Fri, 21 May 2004, Hallam-Baker, Phillip wrote:
Ted's proposal gets each party what they really want.
I agree. Its best to have our own record type in DNS and at the same
time current use of TXT will help in greater deployment for the next
couple years.
While there would be period when some software would prefer query both
records at the same time, I expect in the future we'll move to using only
new records and those systems that query dns will set to only query MARID
records and if they don't get the answer then query TXT (despite that this
will create small delay, if < 10% of records become TXT, it'll be worth it
to introduce small extra delay but cut back on the work of sending extra
packets on the net).
Meng and microsoft get the ability to immediately deploy marid without
waiting for deployment of the extensions handling capability.
This I did not get, dont both SPF and Caller-ID proposals already have
ability to handle extensions? I would expect new proposal would as well.
That does not mean we do not make a good faith effort to help solve the
deployment of dnsext. None of us want to meet this issue again and again.
Perhaps this should be taken as an important points by Microsoft people
here who might perhaps use greater force to convince their peers that this
issue is VERY important and their software needs to be upgraded sooner.
Infact we have a pretty good set of use issues to deal with that should
be captured and fed back. There is still a wildcard issue to be
addressed for srv. I think that can be solved.
I really don't see how. SRV records and any other systems that relies
on adding _???. to main domain is just not workable with wildcards, there
does not seem any way around it other then to make some additional queries
(i.e. by looking at zone listed in authority section and looking there
for wildcard record, etc)
--
William Leibzon
Elan Networks
william(_at_)elan(_dot_)net