An open discussion list for topics related to the eduGAIN interfederation service.

[Ian Young <ian AT iay.org.uk>]

On 27 Mar 2014, at 10:46, Niels van Dijk <niels.vandijk AT surfnet.nl> wrote:

> Naive me was sort of hoping that combining CoC and attribute
> requirements in metadata at some point would lead to something that
> scales better as compared to what we have now.

I think it will, and I think things like the CoC will drive the presence of 
RequestedAttribute a fair bit, but I don't expect any of this to happen 
overnight.

As Leif and Alex have pointed out, though, RequestedAttribute is a pretty 
broken mechanism in various ways. It really wasn't designed for what we're 
trying to do with it. I don't see any perfect solution on the horizon either; 
entity categories may help in many cases, but they are not a direct 
replacement and they won't apply to all SPs. They *do* have the advantage 
that rather than being the SP's shopping list, they can explicitly speak to 
the *purpose* behind the provision of attributes.

> As joining eduGAIN as an
> SP is a deliberate step anyway, I could imagine at that point requiring
> better metadata from the SP.

We don't require it, but we do take the opportunity to push for improvements.

> Should this be be something the eduGAIn community should consider?

You mean, should eduGAIN say that SPs MUST have this metadata? For what it's 
worth, I wouldn't be in favour of that, as I am not in favour of anything 
that raises a bar to participation in a service that is still in its infancy 
in terms of adoption.

> I mean, contacting the SP is always the default option of last resort,
> can't we work towards improving that?

I think contacting the entity's registrar (i.e., the originating federation) 
is the next step up the ladder.

        -- Ian

Attachment: smime.p7s
Description: S/MIME cryptographic signature