I still believe that VRML should be primarily a scene description language
that sticks to describing the object classes and instances making up a
scene as "data" and does not evolve into yet another scripting language.
- Mark Pesce's implementation fits firmly into this model.
For VRML to work, though, given the limited bandwidth of the net, we
need a way to avoid having to send huge geometry files. It is essential
to be able to use scripts to generate scenes locally at the client from
higher level descriptions sent as part of VRML. This is straightforward
to achieve if we use URLs to associate scripts with VRML documents.
I am also very keen to open the flood gates to "ordinary folks" to
create their own rich VR scenes and applications. I want to be able to
use someone else's fractal cloud generator and to combine it with trees
from here and buildings from there. For this to happen we need to keep
VRML simple; to combine it with simple scripting languages; and to allow
the use of geometry data created with commonplace 3D modelling tools.
To encourage the rapid deployment of VRML, we need a libvrml which handles
parsing and provides for integration with OpenGL, PEX and Renderware etc.
I want the library to include a simple scripting language too, to act as
an existance proof of how VRML can support intensional object classes.
It should also support the most common 3D modelling formats such as DXF.
We need to start figuring out ways in which the challenge of building this
library can be shared out across the net. Lets go for it!
-- Best wishes,Dave Raggett
----------------------------------------------------------------------------- Hewlett Packard Laboratories email: dsr@hplb.hpl.hp.com Filton Road tel: +44 272 228046 Stoke Gifford fax: +44 272 228003 Bristol BS12 6QZ United Kingdom