James, thanks for the review.

On 06/06/2013 12:06 AM, Manger, James H wrote:

BEARER tokens dominate OAuth 2 deployments today, but OAuth 2 is deliberately extensible to support other sorts of credentials (eg MAC authentication).

Why is draft-ietf-oauth-dyn-reg hardwired to only support BEARER tokens?

1.3. "Registration Tokens and Credentials" says:

"The Initial Access Token ... is an OAuth 2.0 Bearer Token"


That should probably say OAuth 2.0 Token there -- I see no reason to constrain the Initial Access Token to be a Bearer.

"The Registration Access Token ... is an OAuth 2.0 Bearer Token"


This, however, makes sense -- it's a shared secret between the server and the client for use with registration. Do we have a mechanism for passing dynamically allocated shared secrets around on HTTP calls? Yes: Bearer tokens.

Google's TLS ChannelIDs [draft-balfanz-tls-channelid], for instance, would be a fantastic fit for linking the first registration request with any subsequent registration modifications. The Registration Access Token would be annoying legacy baggage in that situation.


Wouldn't all OAuth tokens be "annoying legacy baggage" then?

It seems that the Registration Access Token is only ever used at a single URI: registration_client_uri. That sounds like the perfect situation to use a "capability URI", effectively putting the token in the URI. Anyone considered doing that? It should significantly simplify the spec.


That would require servers to be able to differentiate clients based on the URI, and we got feedback from people that wanted to have the Client Configuration Endpoint served on a single URI so that the server wouldn't have to track and map different clients to different URI (see Tim's recent comments even). Separating the URI from the Authorization like this, using existing mechanisms, makes that possible.

 -- Justin

--

James Manger



_______________________________________________
OAuth mailing list
OAuth@ietf.org
https://www.ietf.org/mailman/listinfo/oauth

_______________________________________________
OAuth mailing list
OAuth@ietf.org
https://www.ietf.org/mailman/listinfo/oauth

Reply via email to