Seeking a Dao of Groves

[Len Bullard]

Steven R. Newcomb wrote:
> 
> My perspective is that HyTime's main accomplishment, in the
> end, was to make it possible to impose any combination of arbitrary
> structures, rigorously and formally, on any set of components of any
> kinds of information, at any level of granularity.  In effect, to make
> all aspects of all information, everywhere, a single (ginormous)
> queriable database, or to "address any information, anywhere, any
> time, in any convenient terms". 

<snip>

 (Most of the rest of the problem of designing
> HyTime was making sure that all the kinds of imposable structure
> (trees, directed graphs, and schedules) don't trip over each other and
> work nicely together, both semantically and syntactically.)

Then a problem looking for a solution:  In VRML97, we have a monolith 
standard, self-contained, defines its own structures and types.  In 
VRML200x, the scheduled revision, a requirement is to enable multiple 
encodings whose definitional parent is an abstract definition cast 
as an IDL based on the original VRML97 semantics.  The encodings:

1.  The so-named UTF-8 syntax (essentially the original VRML syntax) 
2.  An XML syntax, so named, X3D.

The problem is, VRML97 is rootless.  Its abstract definition of nodes 
and fields where a children field can contain nodes does not map cleanly 
to XML without the use of wrapper elements.  On the surface, it looks
simple: 
map nodes to elements and fields to attributes.  In reality, it doesn't 
work.  Fields can have nodes (type MFNode or SFNode) as members but XML
attributes 
cannot have elements as members.  So in effect, to make the VRML work,
we
create *ugly* XML.  Legal, but ugly.  This ugliness shows up in the DOM 
calls quickly for what are probably obvious reasons.  VRML97 already 
has the 

appearance Appearance { }

and other constructs which are ugly but we are used to them.

Eliot is familiar with VRML97 and the spec is public.  What would the 
groveMasters do with the VRML definition?  I'm not sure much can 
be done to change our course, but at least, this is an opportunity 
to present how what HyTime defines can be applied to a specification 
where XML has failed to solve the dilemmas.

This should concern XMLers.  It will do damage.  XML may not 
be up to the job, abstractly speaking.  It is proof that 
infosets, schemas, DOMs, and so forth, are not enough.

This is also a nice meaty opportunity.  Note that even TimBL has
predicted a 
3D web, and it is emerging.  Economics of VRML aside (there are real 
reasons for the languid state of 3D on the web that have nothing to 
do with VRML97, it takes more platform power so it stays behind 
the text app curve by about a decade), 3D as a host language is looking 
more and more like the next winner of the framework sweepstakes; yet 
the ISO standard may be flawed or at least, ugly where XML is concerned.

> > Authority is choice, and
> > whether as in the Tao, this is opposites, or as in Tao,
> > extremes of the same continuum, intelligence still
> > must choose for community.

> I'll take that as a poem.  (My mysticism parser must be on the fritz.)

Not mystical. Practical as hell;  just obscure.  Why should I change
now? :-)

len


xml-dev: A list for W3C XML Developers. To post, mailto:xml-dev at ic.ac.uk
Archived as: http://www.lists.ic.ac.uk/hypermail/xml-dev/ or CD-ROM/ISBN 981-02-3594-1
Please note: New list subscriptions and unsubscriptions
are  now ***CLOSED*** in preparation for list transfer to OASIS.