Re: [core] FW: New Version Notification for draft-fossati-core-parametrized-cf-00.txt

[Thomas Fossati <[e-mail]>]

Hi Michael,

On Mon, Jun 13, 2022 at 6:24 PM Michael Richardson
<[e-mail]> wrote:
>
>
> Thomas Fossati <[e-mail]> wrote:
>     > On Sat, Jun 11, 2022 at 3:23 PM Michael Richardson
>     > <[e-mail]> wrote:
>     >> okay, so you made me read your draft.
>     >> Conceptually, it seems fine, but do we really need it, is the ask.
>     >>
>     >> There is an assumption you've made that EAT will be carried over CoAP.
>     >> I think that it will be carried over CoAPS, possibly over a DTLS-keyed OSCORE
>     >> or an EDHOC-keyed OSCORE.
>     >> I bring this up because it opens new places in which to specify the expected
>     >> profile.  For instance, ALPN.   Is that a good place? I don't know yet.
>
>     > An interesting idea. If the ALPN'd protocol fully defines the
>     > exchanged media types then it may be all it takes.  Though if that's
>     > the case then one doesn't need content negotiation in the first place.
>
> I think that the Attester just needs to declare eir profile.
> The Verifier either supports it, or it doesn't, and things fail.
> There is really no negotiation possible unless the Attester supports multiple
> profiles, and then we get into questions about bid-down, etc.

I agree.  I'm not arguing that.

>     > The "do we really need it?" is exactly what I'm trying to understand,
>     > but the "we" for me is CoRE users at large rather than the EAT / RATS
>     > populace only.
>
> I don't believe that there are any *CoRE users*
> Many protocols build upon CoAP, CBOR, SenML, etc. but those protocols have names and
> communities that can define a profile.

I didn't want to talk about profiles here :-) I wanted to have a
conversation around extending the content negotiation capabilities of
CoAP, and how the CoRE community (those who own the protocol) feels
about that.

cheers!
-- 
Thomas