[
Lists Home |
Date Index |
Thread Index
]
Ronald Bourret wrote:
>
> james anderson wrote:
> > Is it more of a conceptual challenge to add an attribute declaration
> > to the internal subset than it is to add a global namespace
> > declaration to the internal subset?
>
> Yes, especially since you might need to add more than one.
>
> (Why do you say internal subset here? Validation DTDs are external.)
John Cowan wrote:
>
> james anderson scripsit:
>
> > Is it more of a conceptual challenge to add an attribute declaration to the internal subset than it is to add a global namespace
> > declaration to the internal subset? The only difference is that the attribute declaration rquires that one must determine the
> > "root" element. I would have thought one needed to make that decision in order to encode the internal subset.
>
> I don't understand this talk of internal subsets. I am talking about
> validation DTDs (or "bags of declarations" if you prefer), which are
> external by definition.
I'm uncertain how DSDL wishes to incorporate these DTD's, but had presumed that a document type declaration (that is a DOCTYPE
form) would (at least) not be precluded. Should external subsets only be permitted, then my observation applies to the
"framework" entity which comprises the others by reference. Any "bag of declarations" is going to be encoded in a tree of
physical entities. This "framework" entity is the root of this tree.
>
> > The only difference is that the attribute declaration r[e]quires that
> > one must determine the "root" element.
>
> But this might not be known. If you know what your document looks like,
> it is known, but you can also define vocabularies from which any element
> can be the root (less likely) or from which any element or attribute can
> be reused in another vocabulary (much more likely). The only solution
> for such vocabularies is to add namespace attribute declarations to all
> element type declarations.
>
This last claim is false. One need add declarations for only those elements which appear at the root(s) of a partial order.
These need not be the same elements as the root element.
...
|
