questions re the ChainingMetadataProvider type

[Tom Scavo]

On Mon, May 14, 2018 at 9:52 AM, Cantor, Scott <cantor.2 at osu.edu> wrote:
>
> ... it's not crucial since we don't even intend people wire Chaining up any more.

If that's true, then you may as well deprecate
ChainingMetadataProvider. Regardless, chunks of the documentation need
to be rewritten in that case.

Let me push back on this a little bit. A sequence of providers wrapped
in a ChainingMetadataProvider is a welcome convenience due to the fact
that top-level providers must have a sortKey (if not, the order is
undefined). OTOH, a ChainingMetadataProvider implies the natural
order.

Note that I added some examples to the ChainingMetadataProvider topic.
[1] It's easier to talk about strategy in the presence of a
ChainingMetadataProvider since the order is the natural order. An
explicit sortKey is a nuisance.

Tom

[1] https://wiki.shibboleth.net/confluence/x/NgInAQ