Shibboleth Developers

["Scott Cantor" <>]

> I completed the ARP architecture that I proposed on Mon.
> 
> Please see http://icap.andrew.cmu.edu/aa/AA.htm for the 
> latest version.  I think it can handle most of the issues we talked
> about.
> 
> Please send me your questions and comments.

This makes pretty good sense to me. I think it's consistent with what
we've been trying to sketch out. I think it may be the case (and perhaps
is not really intended to be part of this diagram) that there will be
additional hooks needed by the UI to properly present things.

For example, we've had a notion that being able to assign a recognizable
name to a set of resources would be a good thing, and eventually might
be a way to link up the SHAR and AA in communicating about the resources
in various application domains. But I don't think that's core to the ARP
processing that is being designed here.

Another thought that's more about the implementation is whether it might
sense to think in terms of the ARP:ATTR object "box" actually being an
implementation of the SAMLAttribute interface we've defined, rather than
having it "return" an attribute.

This might be a good thing if we can combine the value filtering
process, the logic that actually has to obtain the attribute values from
their source, and the abstract portion that presents the data to the AA.

-- Scott

------------------------------------------------------mace-shib-design-+
For list utilities, archives, subscribe, unsubscribe, etc. please visit the
ListProc web interface at

    http://archives.internet2.edu/

------------------------------------------------------mace-shib-design--