Laurence, Rohan, Carsten,
Thanks and agreed: a serializations registry belongs to CBOR, not COSE, and
we don't want to slow your serialization draft in any way.

We met (as we are here for 127) and our proposal for -01 is to drop the
COSE registry. The header parameter would carry an identifier defined by
the CBOR WG, not a COSE registry value. "dtrm" looks like the natural
choice.

On Carsten's point about data model subletting, we'll follow the CBOR WG's
steer. New CBOR serialization identifiers would follow your Standards
Action rule.

The parameter itself stays general: it names the deterministic encoding of
the content identified in 259.

-01, which we'll post once open again, defines a value for CBOR content
only. What COSE still needs is the header itself: application/cbor in 259
doesn't say whether a deterministic serialization was applied, so a Hash
Envelope verifier re-encoding the value has to guess.

As Laurence says, that works with or without a registry as long as there's
an identifier.
Please confirm you have no other objections and see you in SF!
Best
Steven and Anton

On Thu, Nov 12, 2026 at 11:46 AM, Carsten Bormann <[email protected]> wrote:

> On 2026 Nov 12, at 20:34, Rohan Mahy <[email protected]> wrote:
>
> new encodings
>
> (For COSE readers who haven’t followed this discussion at the CBOR WG:
> There are no new encoded data items (which could be meant with “encodings”)
> under discussion, just profiles that variously constrain the choice of
> multiple encoded data items that CBOR already provides a data item.)
>
> Grüße, Carsten
>
_______________________________________________
COSE mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to