Re: CR-14 value space of entities, notations, IDREFs

"Henry S. Thompson" wrote:

> To clarify one thing, we consided entities and notations quite
> distinctly from IDREFs -- they're very different cases.
> For entities and notations, another way to put our point is that if
> you _do_ have the components, the difference between name and
> component is trivial, but if you _don't_ it's enormous, so going for
> the name makes sense.

But for ENTITY there isn't any componenent involved. I can see the
argument for NOTATION, but I can't yet see why the value space of ENTITY
should not be entity info items.

> For IDREF, the new draft goes to some length to distinguish between
> type-validity for IDs and IDREFs, and document-validity for documents
> including items validated with these types. It makes the former just a
> matter of NCName lexical form, and only the latter involves notions of
> uniqueness and reference resolution, and is located in Part 1, not
> Part 2.  Accordingly, since it doesn't make sense to include
> reference-dependent information in the definition of the IDREF(S)
> type(s) as such.

That makes sense.

With this clarification, I won't dissent with respect to the WG's
decision on the value space of IDREF.

(Actually I would prefer ID in the Schema to mean "this attribute was
declared as ID in the DTD" (analagously to ENTITY), and IDREF to point
to elements with DTD-declared IDs.  Then the value-space of IDREF could
be element info items with introducing any dependency of Part 2 on Part


Received on Friday, 9 March 2001 07:04:01 UTC