RE: [security-services] Roles for SAML 2.0 Metadata

> I very much agree with Anthony's statement that 'various profiles should
> define the "roles" and not the metadata
> specification'. Indeed, the metadata specification is not seeking to
define
> roles, but rather "identify" existing roles, taken from existing profiles,
> and then determine the metadata that needs to be defined for these roles.

I think it's reasonable to characterize the work item on metadata that way,
but when the editing process takes place, I think a more effective way of
partitioning the material is what I (and I think Tony) suggested; a section
in core for the containing elements and HTTP-based resolution work, and the
rest within B&P to define the actual role descriptors as well as referencing
them within the bindings and profiles.

Probably added to core would also be the basic outline of *how* you define
descriptors (a base element or model group for example to use as a
foundation).

(And yes, I'm willing to help do all that.)

-- Scott