RE: [higgins-dev]http://wiki.eclipse.org/index.php/IdAS_Registries_Propo

["Paul Trevithick" <paul@xxxxxxxxxxxxxxxxx>]

Tom wrote:
> 
> Paul,
> 
> A few questions on your proposal:
> 
> 1. Any XRI with an '=' is always a single DS context, correct?  

As you know, the XRI convention is that XRIs that start with '=' identify
people (vs. '@' for organizations). But we don't know much more than this.
We discover at runtime what services this "person" has. But it's open-ended
what these services might be about. Some of them probably do provide
information about Drummond, the OpenID service type certainly does. The
OpenID service (at present, anyway) only provides information about Drummond
himself (not what Drummond knows about others). So in the OpenID case we
would [use a context provider to] expose this as a single Digital Subject
Context. But that can't be said in general. 

Of course after the IdAS consumer get's their hands on an IContext, they can
inspect whether or not it has (at least as they've authenticated) N>1
Digital Subjects.

> If so, for
> this type, is there an implied relationship between the CID and the one
> CUID that will come from it? 

I don't think there is an implied relationship between the ContextId and the
'SubjectId' [we no longer call them CUIDs; SubjectIds _are_ contextually
unique identifiers].

 I assume there is no implied relationship,
> only potentially a happenstantial one.

Yes.

> 2. How are the CIDs themselves advertised?  That is, as an IdAS consumer,
> how do I find what CIDs are available to my application in order to, say,
> present a set of context's available to users of my application?

Ah, yes. This is a hot topic. To date, we've left the answer to this
question (i.e. "Context Discovery") out of scope for Higgins 1.0. Yet we
know that from so many angles (from enterprise deployment administration to
social networking) answers are needed. I predict many solutions will be
developed outside of the scope of the Higgins project per se. Nevertheless
it is an open question as to whether Context Discovery should be a Higgins
work area. I know of a couple of conversations that suggest that it should
be. And maybe, as seems to be the Higgins default mode of thinking we
conclude that we need a pluggable framework, not a monolithic solution.

> 3. Maybe a half follow on to #2, as an IdAS consumer, how do I know that
> the ContextFactory I resolve from @WISER is actually deployed on the
> system I'm running on?

Good point. I see two possible answers. The simple answer (that works in
simple situations) would be to say that this Context Provider Registry (XRDS
file) is specific to the local system in question. Another (more scalable)
answer is to say that there is a very simple local (third!) registry that
tells what's deployed locally.

What are your thoughts?

> 
> Thanks,
> Tom
> 
> >>> "Paul Trevithick" <paul@xxxxxxxxxxxxxxxxx> 03/01/07 3:41 PM >>>
> Drummond,
> 
> 
> 
> After much discussion with Andy Dale today, this document has been
> significantly revised. Can you please give it a look. Little by little, I
> think we're getting to something workable.
> 
> 
> 
> -Paul
> 
> 
> _______________________________________________
> higgins-dev mailing list
> higgins-dev@xxxxxxxxxxx
> https://dev.eclipse.org/mailman/listinfo/higgins-dev