But you're just pushing the issue up a level, not solving it. We don't
*have* to create a language. We *do* have to create an API, of *some*
sort. (And it's got to be standard across architectures, given how
heterogeneous the Web has gotten.)
My opinion, having looked over a bunch of alternatives, is that the
best option is *probably* creating a hybrid: a simple language that
also acts as an API. (That is, external programs are expected to
generate this language; the language provides the necessary hooks for
editing the scene graph on the fly.) This is the Geomview approach,
and I still haven't seen anything that beats it for
flexibility. Simple applications can work just within this basic
language; more sophisticated ones can use the language interactively
to control the scene graph. Very elegant solution...
-- Justin
Who wonders whether we should be moving
this discussion over to the new (and
so far silent) vrml-behaviors list...
Random Quote du Jour:
"I always thought like Mencken, that if I had been around at the
creation I would have had many helpful suggestions, and when I write
my books on God's Thousand Worst Errors, the first and greatest error
I will charge is that 'He' (or 'She' or 'It') gave us such marvelous
brains and forgot to include the operating manual."
-- RAW