> To sum up, I can always use facets to xsd:ID type and they are honored
> no matter whether the processor supports compatibility datatype or not.
> Am I Right? If so, I'm happy with the outcome.
Right.
> Maybe it's just me, but it sounds inconsistent to me. It says that
> xsd:ID is an alias to comp:ID but actually they differ in parameters.
> That's not what we usually mean by "alias", isn't it.
But the spec doesn't simply say "alias" it says "alias ... for the purposes
of the processing defined in the compatibility spec".
> It sounds more like we have an optional special cross-reference
> semantics, which is applicable to some types (xsd:ID and comp:ID), and
> we have a class of RNG processors that supports this semantics, which we
> call the compatibility processors.
I don't think that's inconsistent with the current specs. The
compatibility spec defines ID compatibility semantics and associates
comp:ID with those semantics. It also defines RELAX NG validation
semantics for comp:ID. The guidelines spec associates the ID compatibility
semantics defined in the compatibility spec with xsd:ID.
I'll see if I can rephrase things to make this clearer.
James
