Re: Why DOCTYPE Declarations for XHTML?

On Tue, 18 Jan 2000, Murray Altheim wrote:
> "W. Eliot Kimber" wrote:
> > Murray Altheim wrote:

> > I'm arguing with you about it because it's important. XHTML will set a
> > precedent, an important one, and it would be a tragedy if it sets the
> > wrong precedent. Pretending that DOCTYPE declarations have any value for
> > defining types is wrong and it shouldn't be done. It seems very clear to
> > me.
> 
> [...] I maintain that within the bounds of our design environment I
> can't see any reasonable alternatives, unless XHTML is take off on
> some entirely new territory, on its own from the rest of the XML
> community.

Well, how about language that makes it clear that this is an open issue?
The prose spec can still mandate the essential requirement(s) - validation
with respect to a specific DTD - with a further explanation of what a
validating processor would have to do (e.g. infer the "right" doctype
declaration - as I've already suggested.)

> > The SGML and XML standards define a universe in which you can only
> > talk about a single document. In that context, DOCTYPE declarations
> > do define types, types with exactly one instance. But that's not
> > very useful.
> 
> [...] the above statement is completely counterintuitive to my
> experience, which is that I've seen *great value* in validating my
> document's markup structure based on DOCTYPEs and DTDs. And continue
> to do so with XML.

I think Eliot means that it's not useful to view a document-specific
definition as a type definition.  It can certainly be useful to repeat the
same definition in many documents - and know/believe that the same
definition is being repeated - which is what I think you're saying.  The
point is that while a true type definition has to exist independent of any
particular instance, there is no way in straight SGML/XML to invoke such a
definition *by type* in an instance.  Thus, the point of a meta-DTD in the
AF formalism, for example, is that such a DTD is not a syntactic component
of an instance, whereas the declaration subset - the effective instance
specific DTD - in a doctype declaration necessarily is.  The two -
definitions for an instance, and definitions for a type - should not be
conflated.  That's not to say that an instance specific conflation can't
be intended, but this *is* a separate (and separable) issue. 

(I'm practically certain that you, specifically, know all this.  I'm
having trouble accepting that The Powers That Be don't understand the
issue: looks more like the PTB are *refusing* to acknowledge it.)

> Since the beginning of XHTML m12n there's been an
> empty XHTML module named "XHTML 1.1 Base Architecture" whose content
> looks like this:
> [...]
> If you look, this is actually included in our current distribution.

A fulltext search for '10744' yielded zilch in these three (single file
version) documents:

http://www.w3.org/TR/xhtml11/xhtml11-20000105.html
http://www.w3.org/TR/xhtml-modularization/xhtml-modularization-20000105.html
http://www.w3.org/TR/xhtml-building/xhtml-building-20000105.html

> If filling this module in, adding it to the DTD, and lobbying for a
> prose section describing what it's for would do any good, I've never
> been opposed to this; I'm the one who snuck the damn thing in. [...]
> Since it's somewhat below the W3C radar, they might simply not care so
> long as they didn't perceive it to damage the spec. 

All hope is not yet lost, then!? :)

> But I honestly can't go in and suggest this as an *alternative* to XML
> validation using DOCTYPE declarations and DTDs, which as I've said
> does hold substantial value for our audience, to whom architectures
> are understood as namespaces and some magic handwaving.

As I suggested, just get rid of #4 in the conformance requirements.  The
part that has crunch is already in #1 and #2.


Arjun

Received on Tuesday, 18 January 2000 23:39:04 UTC