Re: [TECH] Sun's Java and VRML

Paul Burchard ([email protected])
Mon, 27 Mar 95 18:01:41 -0700


[email protected] (Suresh Krishnajois) writes:
> If someone ports something like OpenGL or OpenInventor
> to Java, then will the VRML effort be moot ?

Not at all -- using static formats to describe the initial
configurations of 2D and 3D hypermedia documents still makes perfect
sense.

Although conceptually documents can be composed entirely as a network
of software "objects", authoring "persistent object stores" requires
a lot more software scaffolding than creating simple static ASCII
documents in HTML, VRML, or the like. The simplicity of HTML and
VRML will spawn a larger diversity of tools for manipulating these
formats, and that's a Good Thing.

The model for Web interactivity which I advocate (and hope to propose
in more detail soon) is based on the following relationship between
2D/3D static formats and dynamic scripting languages like Java:

** Scripts are "attached" to static documents. This requires the
static format to be able to embed script references, and to attach
script triggers to suitable format elements (which invoke the script
with given parameters when some user interaction event occurs).

** Static documents are "modelled" by the browser for the script, so
it can act back on them. This requires a standard object model to be
defined for 2D and for 3D documents (document-format-independent at
the highest level, preferably). The browser must then present each
document to the script as being composed according to this object
model (in a scripting-language-independent manner via ILU and/or
CORBA).

The document object models in Java are fairly preliminary, even in
2D, as I think the folks at Sun would agree. VRML fans can advance
the cause of interactivity by designing a good object model for 3D
documents, ready for translation into Java, ILU ISL, and/or CORBA
IDL. I don't know if the models from SGI's OpenInventor would be an
appropriate/reasonable starting point; I welcome your suggestions.

> As I understand it (I may be wrong) the current VRML
> application model is that the application state will be
> stored in the server, which will generate the geometry
> and send this across as an ASCII file to the browser.

That's merely the initial stage -- "behaviors" have always been
understood as essential for VRML in the long run.

--------------------------------------------------------------------
Paul Burchard <[email protected]>
``I'm still learning how to count backwards from infinity...''
--------------------------------------------------------------------