On Thu, Nov 9, 2023 at 4:28 PM George Michaelson <g...@algebras.org> wrote:
> I think we're conflating how you learn what endpoint to send NOTIFY > to, with the protocol extensions or changes to make it legal/normal to > do NOTIFY for this purpose. Agreed. > > I don't personally think the whole "but how do I know where to do it" > is as important as some of you seem to think it is. But, having the > definition of this use of NOTIFY wired down for its protocol elements > and behaviours, I think thats important. I think there is essentially 2 pieces to this which should be different document. 1) how do I find a service provided to the children by the parent, your WHERE, which is probably something that can be agreed upon without too much controversy. 2) the protocol to convey that service, your HOW, which will likely get more diverse opinions as to how to achieve this, and will also need to be more scrutinize. DS update is such protocol amongst possibly many (nameserver update, glue update, ttl updates to name a few). Manu > > > I fully expect the actual endpoint to be OOB. I don't think it has to > be in band but I don't mind if there is an RR which handles in-band, > but to me that isn't the main point: the main point is knowing its a > general expectation I can do this to parent/publisher when I need to, > and what protocol it looks like. > > Maybe I am missing the point. Sure, CDS was all about in-band, but to > me, NOTIFY is mechanistically about a HOW not a WHERE. HOW do I format > records to make a push event happen. WHERE is between me and my > parent. > > -G > > _______________________________________________ > DNSOP mailing list > DNSOP@ietf.org > https://www.ietf.org/mailman/listinfo/dnsop >
_______________________________________________ DNSOP mailing list DNSOP@ietf.org https://www.ietf.org/mailman/listinfo/dnsop
