On Nov 2, 2023, at 23:29, Orie Steele <[email protected]> wrote:
> 
> typ in cose headers is a different draft, it's common in JOSE, and it's 
> recommended in the JWT BCP... I suggest we assume it's not possible to 
> specify the typ (media type) of the envelop in COSE, since that's true today.

Obviously we could register it with ccs/cwt.

> My point is that security assumptions are the same for both the protected 
> header and the protected payload, as are the semantics for "claims", in the 
> context of CWT... At least that's what the current draft says.

They are secured the same way.
That doesn’t mean what they say has the same semantics.
The cross-protocol issue I’m chasing after can exploit the undefinedness of the 
latter.

Grüße, Carsten

> I don't think this draft should say anything more.
> 
> OS
> 
> 
> 
> 
> On Thu, Nov 2, 2023, 5:11 PM Carsten Bormann <[email protected]> wrote:
> On Nov 2, 2023, at 23:02, Orie Steele <[email protected]> wrote:
> > 
> > I suggest we tackle these issues in a separate document.
> 
> I’m fine with that, as long as that document can make retroactive BCP14 
> statements :-) (*)
> 
> The CCS in the payload is entirely different from one in the header:
> The CCS in the payload is the focus of the signed/encrypted/mac'ed statement.
> The CCS/CWT in the header can only be supplementary information to what is in 
> the payload.
> How does that supplementing affect the entire construct?
> Mike proposed using typ to supply this information.  But then it really needs 
> to.
> 
> Grüße, Carsten
> 
> (*) OK, there is precedence in RFC 8725
> 
> -- 
> last-call mailing list
> [email protected]
> https://www.ietf.org/mailman/listinfo/last-call

_______________________________________________
COSE mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/cose

Reply via email to