Re: On the VRML 1.0 spec, and VRML in general

James C Deikun (
Sat, 10 Dec 1994 02:53:41 -0500 (EST)

On Fri, 9 Dec 1994, Brian Behlendorf wrote:

> On Wed, 7 Dec 1994, James C Deikun wrote:
> > Not even HTML lives by the static alone--see forms, imagemaps, and cgi
> > scripts of all kinds. Clumsy, but they show the desire and need for
> > interaction beyond the pure static model. The demands for interactivity
> > in 3d scenes will be much greater.
> Most definitely, but I don't see what VRML has that disallows this
> either... once you read a vrml file into an application that application
> can do whatever it wants with it. What you want is some way of
> controlling that application's interaction with the file, or
> communicating interaction over the net - the concensus was, let's wait to
> see how the static language is used in a dynamic environment before
> deciding upon a dynamic language that might not fit the bill exactly
> right. Open Inventor has shown there is at least one path for this to
> take.

Well, yeah, what I was trying to agitate for was a standard format for
snippets expressing modification of a vrml world that's already been read
in. I don't think this is something that you can seriously put off--this
sort of thing is going to be absolutely integral to a wide spectrum of
vrml apps, and in particular those with the most 'wow' value. If this
need is not taken into consideration and provided for at every step of
the standards process, you'll wind up with browser writers doing it for
you, badly, just like HTML inlined images and forms. Like the html wg
you'll be hard pressed even to keep up in describing what's going on,
much less to actually manage to fix all the things that are broken about
the de facto standard.

The way I see it, it'd be good to be able to define nodes with
substitutable parameters (you'd need this for animation anyway) or maybe
just have a syntax for specifying a particular field (or list element in
a field) of a node to replace. This would work by starting at a named
node and descending recursively guided by field names and indices. The
slot specified can then be filled with a node in regular vrml syntax
(plain node, or DEF or USE). If, however, one specified just a name
rather than an index or field spec the name would be redefined. This is
just a rough first cut, of course, and likely a lot could be done for
efficiency, but it gets across a good approximation of the functionality
that will probably be needed for maintaining dynamic scenes...

James Deikun (University of Pittsburgh)           ...and his trained .sig
#include <std_disclaimer.h>                     // my opinions are my own
logomachy - a fight about words     |     windomachy - a fight about OSes