From: roskos@telerobo.rutgers.edu (Roskos)
Subject: Re: TECH/PHIL: World Descriptions
Date: 3 Jul 92 06:42:59 GMT
Organization: Rutgers Univ., New Brunswick, N.J.



[ a great deal of stuff deleted ... ]

Well, there certainly were a great many questions asked.  But many of
the questions inter-relate to one another, and a solution to one would
imply an approach to solving the others.  I have recently addressed
many of these questions recently.  The ultimate answer I derived was
that there is no single best answer to how a virtual reality simulation
should be set up.  For instance, some simulations would benefit from
centralization (of control and data), and some wouldn't.  In fact, a
variable degree of centralization is the best.  This centralization
should be divided along types of data and types of control.  Objects
should tend towards being more generic than specific since no one wants
to write new code for each new object.  A mixture of TCP and UDP
connections should be implemented for network communications as no one
is necessarily the best to use.

I have submitted an article for publication in the Journal of Parallel
and Distributed Computing, Special Issue on simulation, in which I
outlined a description language specifically to describe a virtual
world.  The language addresses machine and device availability and
allocation.  Simulation-time management of devices and
object-to-machine bindings is, however, not as easy to describe.  As I
have not received feedback yet from the editors of the journal, I am
hesitant about distributing any copies.  But I would enjoy and
participate in discussions about these topics in this newsgroup.