Re: [Cbor] Unusual map labels, dCBOR and interop

[Wolf McNally <[e-mail]>]

Orie,

> On Mar 28, 2024, at 11:01 AM, Orie Steele <[e-mail]> wrote:
> 
> <no hat>
> 
> Sounds like the advice is:
> 
> For more specific media types such as gordian envelope, make it clear that while they might end in +cbor they are actually a subset of cbor / dcbor.
> As you have said this is essentially the same as what we do today, when creating more specific media types for cbor based application formats.
> This advice belongs in the dcbor draft, but a new media type for generic "application/dcbor+cbor" is maybe not needed since we don't expect any protocols to use generic "application/dcbor+cbor" or really even "application/cbor" when they have more specific media types available such as:
> 
> application/cose (which is never dcbor)
> application/cwt (which is never dcbor)
> ... https://www.iana.org/assignments/media-types/media-types.xhtml (anything with +cbor, is only dcbor when it says its dcbor)
> 
> ... and it's forbidden to update a media type that was previously cbor to dbor because doing so would create a breaking change.
> 
> I think this advice is worth writing down.
> 
> OS

Since registering an `application/<x>` media type requires RFC status, we have held off on proposing specific media types for either dCBOR or Gordian Envelope until such a time as the community feels that the time is right. Even then, despite the fact that it would be semantically correct, intuitively we’d like to avoid unwieldy constructions like `application/envelope+dcbor+cde+cbor`.

This is definitely a question open for discussion.

~ Wolf