Re: [Dance] draft-ietf-dance-client-auth-01 sec 8

[Robert Moskowitz <[e-mail]>]

Shumon,

In DRIP, there is no plan to mandate X.509 certificates behind the 
keypair forming the DETs (see rfc 9374).  There will be some optional 
X.509 defined at some point.

The focus right now is on the DNS that will provide access to the SPKI 
for the keys used in DRIP.  This DNS will provide for both TLSA and HIP 
support.  There is also a RATS-style Endorsement of the DET/PK that will 
also be in DNS (for starters a CERT RR with private OID, perhaps a new 
RR or new type for TLSA).

Thus YES it is DNS for the SPKI for the client as well as the server.  
Of course in some cases like Detect And Avoid (DAA) who is on first is 
open to events in the sky.  In Network Remote ID (see 
draft-moskowitz-secure-nrid-c2) the client/server roles are clear.

And since the SPKI in TLSA will be available, hashing it makes no sense.

On 5/4/23 22:26, Shumon Huque wrote:
> On Thu, May 4, 2023 at 9:50 AM Robert Moskowitz 
> <[e-mail]> wrote:
>
>     Raw Public Keys
>
>     Should Matching Type Field of '0' be included here that then it is
>
>     3 1 0 (SPKI)
>
>
> Robert,
>
> I don't see a specific reason to restrict raw public keys to only 
> matching-type 0, unless your application is also using the DNS 
> to discover the client public keys.
>
> Can you elaborate on your rationale?
>
>     Since I am using EdDSA25519, the SPKI is 44 bytes.
>
>
> I guess, matching type 2 (SHA512) makes less sense in that case since 
> the data field part of the TLSA record will be larger than that.
>
> Shumon.
>
>