[Cbor] Re: WGLC redux (Re: 🔔 WGLC on draft-ietf-cbor-cde-08)

[Carsten Bormann <[e-mail]>]

> On 2025-03-31, at 16:35, Anders Rundgren <[e-mail]> wrote:
> 
> On 2025-03-31 16:00, Carsten Bormann wrote:
>> On 2025-03-31, at 14:57, Anders Rundgren <[e-mail]> wrote:
>>> 
>>> We can WGLC cbor-cde anytime, personally, I stick to RFC8949.
>> Well, CDE sticks to RFC 8949 as well.
>> I don’t think you win anything by acting otherwise.
> 
> A difference is that my take on the matter presumes a profiled representation/encoding scheme, including for numbers.

I don’t know what “profiled” means.
CDE defines (uses RFC 8949 to nail down) an encoding scheme for numbers.
This is exactly the one you use as well, so you might as well admit that your document uses CDE.

CDE doesn’t tell your application how to *use* numbers (is a temperature in Fahrenheit?  Kelvin?  An integer giving centi-Kelvins?).
I don’t think your document does, either.

> Since the latter has proved to be impossible to reach an agreement on,

Of course there is no agreement on Fahrenheit vs. Celsius :-)
I think you are alluding to the existence of dCBOR.
Nobody has to use dCBOR (even to get deterministic representation), but you object to its mere existence.
That’s weird.

> this work (regardless if it is good or bad), is clearly out of scope for the IETF.

At this point I have lost what “it” is.

> Since it nowadays supports 99.9% of the CBOR constructs

Ah, it = Anders-base, because CDE supports 100 % of the CBOR data items.
I do have some sympathy with not implementing non-zero payloads for NaNs, as few applications need it, so there really isn’t much difference in practice — only for applications that need them.
CDE is there for those applications, too.

(For some reason, your current text excludes “signaling” NaNs only.)

> being candidates for deterministic encoding (only NaN with payloads didn't make it to the party), I'm thinking of rebranding this work-item to "CBOR Core".

A better name would be “CBOR fringe”, as the choice for deterministic encoding in the first place is likely continuing to be a minority thing (which doesn’t mean it’s wrong — it’s OK to bring a chainsaw to the felling of a tree, but maybe not to the buttering of a sandwich).

Grüße, Carsten