Re: [xri] Redirect & Ref Processing Revision #5

Drummond Reed wrote:
> I just posted Revision #5 of the Redirect & Reference Processing  
> proposal at:
>
>             http://wiki.oasis-open.org/xri/XriCd02/RefProcessing
>
> This is the version we will go over on tomorrow’s call.
It all makes sense to me as it is on wiki now, in a mechanical way.  
But I still have some huge questions about how it operates.

I've re-attached Andy's excellent explanation of the use case for XRD  
level Refs, in case anyone missed it. Looking on the wiki at the "Ref  
Example #3: At the XRD Level", which is the one that addresses Andy's  
use case:

Suppose that instead of one Ref at the XRD level, it had two - the  
second to @foo*bar. And suppose that -both- @x*y and @foo*bar offered  
OpenID SEPs.
1. There has to be a clear precedence rule, so that different OpenID  
SEPs don't get selected randomly.
2. If the SEP at @x*y is selected, is the identifier presented to the  
OpenID service @x*y or is it @a ? It seems to me that to keep these  
aggregations of services composable, it would have to be @x*y that is  
presented, i.e. the OpenID server at @x*y shouldn't have to know  
anything about @a. So I have switched identifiers and CanonicalIDs --  
I arrived at an RP and entered "@a" in its OpenID login box, but I  
have authenticated as @x*y. I suppose in some sense this is like an  
OpenID delegate URI, but will the OpenID servers understand it that  
way? Just asking  ;-)


=vg