Re: [Rats] Where does a EAT end?

Inline…

From: RATS <rats-bounces@ietf.org> on behalf of Thomas Fossati <Thomas.Fossati@arm.com>
Date: Sunday, June 5, 2022 at 11:48 AM
To: Carl Wallace <carl@redhoundsoftware.com>, "Smith, Ned" <ned.smith@intel.com>, Laurence Lundblade <lgl@island-resort.com>
Cc: Giridhar Mandyam <mandyam@qti.qualcomm.com>, "rats@ietf.org" <rats@ietf.org>
Subject: Re: [Rats] Where does a EAT end? - consensus?

Hi Carl,

> Carl Wallace <carl@redhoundsoftware.com> wrote:

>  

> Laurence's one-line PR is roughly equivalent to RFC8126 "standards

> action" policy, but without a registry - which I am quite convinced is

> superfluous in this case.

>  

> [CW] Why would a registry be superfluous? All four of the options in

> RFC8126 section 4.9 have registries.

The CDDL that plugs into the socket requires no codepoint allocation,

only a document that contains and describes it.  There are likely other

registries impacted (e.g., CBOR tags, Media Types, CoAP Content Formats,

etc.) but they are modified by IANA actions defined in the RFC that also

provide the extension CDDL.  There is no need for a "synoptical"

registry I think.

[CW] OK, it may be that what I was thinking needed a registry (i.e., tagged structures) is already covered elsewhere.

> PS: we could add a postil "[...] MUST be defined in a IETF standards

> track document _that updates this document_" to make an explicit

> back-pointer in the RFC series.

>  

> [CW] The changes in the current section 1.2 over the past year and the

> recent addition of the collections spec suggests things may still be

> moving. Even if no registry, it seems premature to preclude use of the

> experimental track.

Given that the goal is to maximise the chance of successful interop,

what is the impact of also allowing experimental (alongside standards

track) in terms of implementation support?

For full disclosure and with my Arm hat on: it'd be bad if a

collection-based EAT wasn't supported by most of the ecosystem only

because it ends up being defined in an RFC tagged "experimental."

[CW] I wasn’t suggesting we’d want “experimental” RFCs in the long run, but was only observing the pace of change in this area didn’t seem to fit with proclaiming it “finished”. Use of extensibility mechanisms has not tended to update core specs like this for the specs I’ve generally deal with (things like CMS). We don’t need to belabor this though, if a registry is something the seems useful later, it could be added later.

Cheers, thanks,

IMPORTANT NOTICE: The contents of this email and any attachments are confidential and may also be privileged. If you are not the intended recipient, please notify the sender immediately and do not disclose the contents to any other person, use it for any purpose, or store or copy the information in any medium. Thank you. 

_______________________________________________ RATS mailing list RATS@ietf.org https://www.ietf.org/mailman/listinfo/rats 