Would it make sense to support two scenarios? (1) Discovery as described in my 
original posting independent of "functional" requests. (2) Discovery for 
unauthorized requests (WWW-Authenticate header).

The later might be a lightweight variant  of the first scenario.

regards,
Torsten.



Am 02.08.2010 um 22:20 schrieb Torsten Lodderstedt <tors...@lodderstedt.net>:

> In the WG meeting at Maastricht, I have been asked to write down my 
> requirements regarding Discovery. And here they are:
> 
> Assumptions: Discovery allows a compliant client to automatically obtain all 
> deployment specific data required to securely access a particular resource 
> servers as well as its respective authorization server for the purpose of 
> obtaining access tokens.
> 
> Starting point of the discovery process is a resource URL. This URL can be 
> hard-coded into the client, bundled with the applications resources or 
> manually configured by the end-user at runtime.
> 
> 1) Client -> Resource
> The client uses the resource URL to obtain resource-specific data, such as
> a) one URI refering to its authorization server
> b) a definition of all scopes of the resource
> c) additional data, e.g. supported signature methods and algorithms
> 
> I do not make any assumption about how many requests are processed in order 
> to gather this information and whether there will be any levels of 
> indirections (e.g. from resource to resource server).
> 
> 2) Client -> Authz Server
> The authz server URI obtained in step 1 is used to gather the following data 
> from the authz server:
> a) endpoint URLs (end-user authorization, tokens, ...)
> b) information about the server's capabilities, e.g. the supported response 
> (end-user authorization endpoint) and grant types (tokens endpoint)
> 
> The client stores the authz server's discovery URL along with the token(s) it 
> obtains for further use.
> 
> And that's it from my point of view. I would very much appreciate if we could 
> have a discussion towards a consensus about discovery requirements.
> 
> regards,
> Torsten.
> 
> 
> 
> _______________________________________________
> 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