> -----Original Message-----
> From: Brian Eaton [mailto:[email protected]]
> Sent: Monday, January 10, 2011 3:25 PM
> To: Eran Hammer-Lahav
> Cc: OAuth WG
> Subject: Re: [OAUTH-WG] Proposal to drop/relocate
> response_type=code_and_token
> 
> On Mon, Jan 10, 2011 at 3:06 PM, Eran Hammer-Lahav
> <[email protected]> wrote:
> > What about the difference between the two access tokens? The one
> issued directly and the one via the code? Are those the same? Same scope?
> Same duration?
> 
> Same.
> 
> > I think this needs to be presented as a separate profile from the user-agent
> one because it will make it easier to better describe the security
> consideration of each.
> 
> That seems wrong, AFAICT everyone interested in implementing the user-
> agent profile supported the mode where a verification code is returned.

Supported as in "+1" or running code?

These are two very different client profiles. In one, the client is completely 
authenticated, residing solely in the user-agent. The other is a mix 
authenticated and unauthenticated, where parts of the client can keep a secrets 
but others can't.

Being able to keep a secret is the primary differentiator when picking which 
profile to use. I agree that the hybrid one is an optimization of the 
web-server profile (removing the need to wait for the server to send a token 
back to the user-agent after exchanging the code). But that only means the 
code_and_token really belongs with the web-server profile than with the 
user-agent.

Moving to the profiles specification organization means that we need to present 
each distinct profile separately, and the code_and_token is clearly a distinct 
profile.

With your clarifications, I feel comfortable leaving it in, but not as a hack 
of either one of the other two profiles.

EHL
_______________________________________________
OAuth mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/oauth

Reply via email to