RE: [TECH] Sun's Java and VRML

Jan Hardenbergh (jch@nell.oki.com)
Mon, 27 Mar 95 19:37:00 E


From: sujo@netcom.com (Suresh Krishnajois)

> When I first saw Sun's HotJava (nee Webrunner) 6 months ago when I was
> at Sun, it was like being in a seance =:). I saw tons of things it could
> do for networked VR, but unfortunately couldnt talk about it outside
> of Sun :( But now that it's been released for the masses, I wanted to
> get a thread going here on the HotJava browser and Java
language/technology,
> (URL: http://java.sun.com) and its relevance to VRML, especially in
> context of the discussion we recently had about implementing behaviour,
> and also the discussion a long while ago on procedural vs. declarative
> paradigms.

Thanx for the pointer on Java. As for the behavior discussion, we also said
we would hold off until VRML 1.0 was declared done.

> If someone ports something like OpenGL or OpenInventor to Java, then
> will the VRML effort be moot ?

I can't play with the 3D applet - no Sun, but reads in Wavefront .obj files.
The language has no inherent support for 3D graphics. They will probably
rewrite that example to use VRML - just a guess. When they add 3D to
the language, then it may be moot. And, of course, the market makes up its
own mind about what is moot.

> 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.
> But with HotJava (which can store and execute arbitrary code), only
> application state transitions need to be propagated across the network.
> Geometry can be completely the browser's lookout. Since application state
> is a higher level abstraction compared to geometry, the HotJava model
> will be less bandwidth intensive than the VRML model.

Yes, you are wrong. The ASCII text file (compressed, usually :-) is a
scene/object description. So, the geometry is owned by the browser.

Propagating the "application state" is interactivity. We're holding off on
that.

> Also, if mere application state is what is going across the pipe, and
> geometry is entirely upto the browser, then I can extremely easily
> implement scenarious involving what could be called "camouflage/facemask"

> transformations - kind of like a visual analog of anonymous remailers.
> As Cyberspace becomes more visual and interactive, this capability
> is going to be _extremely_ important. I cannot easily do this if I'm
> constrained to communicate at the level of application geometry.

I think it will be possible to add some serious senarious scenarios to VRML
by means of behaviors without adding a superset of C++ to VRML. However,
a nice general purpose language running through the web would be neat.
Also check out SafeTCL and hopefully Telescript will get out to play.

When we start the discussion on interactivity & behaviors, we should
consider
all the options.

> Please feedback, flame, throwup, whatever on this and related issues.

verbing the noun feedback makes me think of HURL (Horrible URLs)

YON, jch@oki.com, Jan C. Hardenbergh, Oki Advanced Products 508-460-8655
http://www.oki.com/people/jch/ =|= 100 Nickerson Rd. Marlborough, MA 01776
Imagination is more important than knowledge - Albert Einstein (1879-1955)