Re: Update on namespaces

> From: Martin Bryan <mtbryan@sgml.u-net.com>

> At 21:37 11/6/97 +1000, Rick Jelliffe wrote:

> Why not data attributes with an XML-USES attribute?

In addition, sure. 

I am proposing that perhaps:

* entities should be able to have notations (i.e. an XML-USES data attribute)
* elements should be able to have notations (i.e. an XML-USES attribute, does lextyping)
* notations should be able to have notations (cascading notations: requires 8879 enhancement:
  my examples used keyword "USES")
* PIs should be able to have notations, as their first token (i.e. formal processing instructions)
* documents should be able to have notations (i.e. the SEEALSO parameter), e.g.

<!DOCTYPE x SYSTEM "x.dtd"
	SEEALSO 	PUBLIC "IDN//w3.org//NOTATION xml-lang//EN" 
			PUBLIC "IDN//w3.org//NOTATION xml-link//EN" 
			PUBLIC "IDN//w3.org//NOTATION xml-style//EN"
			PUBLIC "IDN//sgmlopen.org/NOTATION CALS table model//EN"  []>


> What changes other than the new TC would be required?

New keyword USES on notations.  I think also remove any restrictions on what things can have
notations. 

> > a SEEALSO 
> >parameter on the DOCTYPE declaration, to declare notations that together
> convert the
> >element set declarations into a DTD (document type definition). E.g:
> 
> What does this mean?

The DOCTYPE declaration & prolog give you element sets. A full document type definition (DTD)
includes more than this: documents have other constraints and conventions and embedded languages.  The SEEALSO parameter allows the
extra constraints and conventions
to be declared in the document, by identifying public or system identifiers. The public text
pointed to by these identifiers could just be plain text descriptions of what is needed. 
Any software functionality for that notation should be keyed by the identifier string, not
by examining its actual text.

Thus, the DOCTYPE declaration can become more like a full  DTD.  In the TC, the SEEALSO
(additional requirements) parameter is part of the SGML declaration. I think this is wrong, 
and it would be more correct to put it in the DOCTYPE declaration, as in the example above.


Rick Jelliffe

Received on Wednesday, 11 June 1997 13:50:38 UTC