Re: [dnsext] WGLC: RFC6195bis IANA guidance

[Mark Andrews <[e-mail]>]

In message <[e-mail]>, Paul Hoffman writes:
> On Jul 3, 2012, at 7:13 AM, Mark Andrews wrote:
> 
> >>>   RRTYPE assignments need to say something about the rdata format
> >>>   no longer being subject to change.  Its not for reserving a code
> >>>   point as it will be coded into implementations.
> >=20
> > DANE thought that they had a code point that they could change the
> > internal structure about at will by going through the allocation
> > process.  They needed to be disavowed of this notion.  This should
> > not happen again.
> 
> Mark's sneering is one way to look at it.

For a expert to be able to evaluate a RR type definition it needs
to be stable.  No expert can say a RR type definintion can be said
to be safe / good if it is still subject to change.

> Another is that, with an =
> RRtype in a mature draft that labelled as TBD, implementers will just =
> pick one and then eventually squat on it, screwing up both the registry =
> and implementations.

Then pick a private type for testing.  Been there, done that.  It is
not impossible to do.

> Standards-track protocols can be changed all the way through IESG =
> review. Those changes could in fact change the internal structure of the =
> data covered in an RRtype. This leaves exactly two choices:

And that ends up with a total review of the impact of the new
structure has on the old implementations.  This level of review
doesn't happen between I-D revisions.

> - Force Internet-Drafts to keep their RRtype as TBD all the way until =
> the IESG passes the Internet-Draft to the RFC Editor
> 
> - Allow WGs and document authors to request RRtype assignment earlier in =
> the process and know that there might be a change based on technical =
> review of the document
> 
> Pick one.
> 
> --Paul Hoffman=
-- 
Mark Andrews, ISC
1 Seymour St., Dundas Valley, NSW 2117, Australia
PHONE: [phone]                 INTERNET: [e-mail]