Re: [DNSOP] HTTPS SVCB no service available signal.

[Joe Abley <[e-mail]>]

Hi Mark,

On 10 Jul 2020, at 21:21, Mark Andrews <[e-mail]> wrote:

> Joe is trying to CHANGE the specification of MNAME and I’m pointing out some of the potential uses of the existing specification.

I think what I am actually trying to do is propose a value for the MNAME field in cases where there is otherwise no clear value to use. Whether we call this a change (or a CHANGE :-) or a clarification seems like a weird thing to fixate on.

I would like to understand what the existing use cases are that you mention.

In the case where there is no primary master in the sense used in 2136 or 1035, what are the potential uses of that field, and which of those uses would be defeated by the use of a standard value (e.g. the empty label) to indicate that there is no useful data included there? Common alternatives are to arbitrarily copy RDATA from an apex NS RR, or to make something up that may or may not be an actual host name. Those alternatives seem to have less diagnostic value, to me; a value that means "nothing to see here" is surely a better signal to someone trying to troubleshoot.

I appreciate this sub-thread has nothing much to do with SCVB (and sorry for overloading), but your enthusiastic objection to the idea of codifying the use of the empty label more broadly than just with that RR has intrigued me.


Joe