I don't understand where this fear is coming from. How do you plan
to wed Java and VRML without designing an OO API? Once you have an
OO API, what's so frightening about expressing it in IDL and
thereby making it (potentially) available to multiple languages?
I'm as big a fan of Java as anyone, and I certainly think we should
aggressively work to bring the two together. I even agree that
picking one initial "preserred" language for VRML is an excellent
strategic move. But I don't see the urgency -- or purpose -- of
permanently locking VRML into a single scripting language.
The *hard* part of standardizing VRML 2.0 will be designing and
agreeing on a good OO API. All I'm asking is that the VRML 2.0
standard be based on the essentials -- the abstract API we come up
with -- and not superficialities of a concrete Java binding.
I'd also like to point out that this is a very difserent situation
(and a much bigger gamble) than our earlier standardization on
Inventor syntax. SGI made QvLib freely available to other
developers, helping them implement "Inventor emulation" relatively
quickly. Java is a *much* more complex system, and Sun clearly does
not plan to give away its compiler or euntime source code to
commercial developers. How does your proposal not amount to
requiring commercial developers to shell out $100,000 or whatever
Sun's license fee happens to be?
Again, I'm all for using Java as an express train to the future.
But let's not handcuff ourselves to the seats, or we'll miss our
station.
--------------------------------------------------------------------
Paul Burchard <burchard@cs.princeton.edu>
``I'm still learning how to count backwards from infinity...''
--------------------------------------------------------------------