Re: OMITTAG specifications in element declarations

DAVEP@acm.org wrote:

> <jenglish@crl.com> (Joe English) recently wrote:
> >Since most SGML users will have to rewrite their DTDs
> >to make them XML-able anyway, leaving out the omitted
> >tag minimization parameter doesn't seem like that big
> >a deal...
>
> But rewriting them so that the content models obey the
> restrictions is a one-shot thing,

Don't forget: you also have to expand all the parameter
entity references, replace things like:

	<!ELEMENT (a|b|c) - - (long, hairy, model)>

with

	<!ELEMENT a (long, hairy, model)>
	<!ELEMENT b (long, hairy, model)>
	<!ELEMENT c (long, hairy, model)>

and change things like:

	<!ELEMENT x - - (a & b & c)>

into:

	<!ELEMENT x ( (a, ((b,c) | (c,b)))
		    | (b, ((a,c) | (c,a)))
		    | (c, ((a,b) | (b,a))) )  >


Heaven help you if you're using something like
TEI, DocBook, or IBM ID Doc...


> and then you can
> continue to deal with those documents using the new
> DTD with its new content models in both SGML and XML
> systems.

I for one do not relish the thought of maintaining
even a moderately large DTD after expanding all
the parameter entities...

> The point of Eve's proposal is to avoid having to
> continue to maintain both versions.  That to me
> _does_ "seem like that big a deal...".

Point taken, but I think it will take more than just
allowing "- -" in the XML grammar to make this feasible.
I could live without AND groups, but parameter entities and
the ability to specify more than one associated element
type in <!ELEMENT ...> and <!ATTLIST ...> declarations are
essential.


--Joe English

  jenglish@crl.com

Received on Saturday, 2 November 1996 20:38:18 UTC