RE: Linked XRD Processing Rules

[Eran Hammer-Lahav <eran@hueniverse.com>]

> - Is there a need for general purpose processing rules to link one XRD
> to other XRDs?

I am not sure. It seems like the only real use case we have is host-meta, which bring back our discussion on the 'match' attribute. There are obviously going to be protocols with a need for this, but do we have actual examples other than the current need to delegate host-meta? 

The need for #replace is completely trust related, because without trust (or with the use of HTTPS), HTTP 3xx covers this (and at the moment HTTP is the only transport we have use cases for).

If host-meta solved its delegation needs some other way, neither will be needed.

So with the information I have today, I would say no.

> - If yes, is a #replace directive enough, or do we need #include?

> - If yes to #include, do we need it today or can we defer it to another
> spec with more advanced processing rules protocols can use?

EHL