[abfab] Direction Forward for aaa-saml

[Sam Hartman <[e-mail]>]

Leif, I wanted to write up my understanding of our proposed direction
forward, just in case things take longer than we anticipate  to
implement.

During the meeting  Monday we discussed Alejandro's proposal.

He proposes adding two new role descriptor subtypes: Radiusidpdescriptor
and Radiusrpdescriptor.
That seems great.

He proposes adding a RadiusIdpService and RadiusRpService of
EndpointType as well.

In the meeting we discussed that we really aren't specifying an
edpoint.  In particular, the location of the service is implicit in this
RADIUS binding.  It's possible we might describe something in the future
where we included radsec endpoints and keys in metadata, but that's not
what we need now.

However we also discovered that a role descriptor doesn't actually need
any EndpointType subclasses.  So, instead, Alejandro will create a
different extension to rolldescriptor to include the naming information
we need.

This will allow us to avoid registering an unresolvable URI to describe
the security name of a RADIUS entity.

Have I accurately summarized what we discussed?
If so, I'd like to solicit any comments from the list.

--Sam