Just an interesting observation: as was pointed out a few messages sooner
(I forgot by whom), SGML (at least, good SGML) *is* hierarchically structured,
just as VRML, that is why VRML syntax should be isomorphically expressable
in a SGML DTD. But: There would perhaps nothing in the document *besides*
markup, nothing equivalent to the text. From SGML's original intention,
this may be odd, and without text it is perhaps not worth to bother
with the markup delimiter syntax, which is why the concrete VRML syntax
doesn't correspond easily to an SGML delimiter set.
Then one could put text among the markup, which would then additional
information to the concrete VR scene, which is expressed by the element
structure and the elements' attributes. Hence, to hierarchically
describe a VR scene in a markup kind of way would imply to express
the metainformation of that scene in the text, not the markup. This
is quite contrary to what happens with ``normal'' text.
Funny? Was I clear at this...? Any inspiration to draw from this?
--Steffen Meschkat mesch@artcom.de