Re: [e2md] Problem statement, Requirements and Use Cases (was RFC 5507 & RFC 5395)

[Dean Willis <[e-mail]>]

PFAUTZ, PENN L (ATTCORP) wrote:
> So I went through the RFCs which seemed to say, "You probably should
> use a new RR type if you're going to get too much data and it's not
> as bad getting new RRs as you think."
> 
> Naïve question: if we said we wanted a WG based on the assumption
> that we'll define new RR type(s) as needed would that settle the
> naysayers' hash?

I believe it would settle some of their hash. Not all of it, of course;
if we propose an RR-type for putting a caller-ID photo jpg into the DNS,
they're still going to get upset and say "no". And preserving their
ability to say "no" on a per-case basis is something they really want.

However, if we agree to, for each use case, consider whether we should
re-use an existing RR type or define a new RR-type, and that we will
also make a similar determination about whether to put the data itself
into the DNS or put a pointer to the data into the DNS, then I think
we're going to get a lot less pushback than our current proposal "enjoys".

I had a side discussion with Jim about his experience in getting new
RR-types approved, and to summarize, my impression was that it was kind
of random. Some RR-types were unexpectedly easy, others unexpectedly
difficult.


--
Dean