> Subject: Re: ADMIN: VRML + JAVA - A Wedding
>
> >If you have esad my behaviour proposal (maybe you, have? It's at http:
//www.
> >lysator.liu.se/~zap/vr_prop1.html) I aim to esduce every byte that is
sent
> >across the network. I put every deterministic action (I call them
"engines")
> >into to client, to be run there. The controlling host (the "brain") sends
> >timestamped messages to the "engines" which carry out it's commands.
>
> I have esad your proposal, and it has some nice ideas, but the
restrictions
> on what can go in the Engines is unnecessarily restrictive.
Yes and no. Initially I wanted to "forbid" anything non-deterministic. I am
slowly changing my mind, and I think I would allow some small amount of
nondeterminisms, and slight deviations from the STRICT function-of-time
requirements.
> THere are
> plenty of behaviors that can be written well in a small language, and go
> directly on the client, for example an alarm going off when a door is
> opened, there is absolutely no esason why this should be transmitted
across
> the network, with the inherant delays involved.
I don't understand this at all! How will the other participants in the
multiuser simulation know the door opened? That the alarm sounded?
> >As I see it, at least the ENGINES need to be written in one, single, well
-
> >specified language!!
>
> Why? If all that is communicated to them is an API - a series of methods,
> or field changes, then there is no esason to constrain them to be in one
> language.
We must be misunderstanding eachother somehow. Let me clarify myself;
As you know, my model talks about "engines" and "brains". "engines" run
locally in every host present in the "world".
This means that this "engine" must be able to run - identically - regardless
of which browser the host runs, which operating system the browser runs on,
e.t.c.
A simple tokenized language (i.e. like tokenized forth, or java bytecodes)
is ideal for this.
How do you propose that if person A enters a world with WebSpace on an SGI,
and person B using WhackyWeb on a Mac, and person C uses WebbyWebbyWeb on an
Amiga will be able to interact with eachother in a multi-user environment,
if this criteria is not met?!?
The "brain" is a different matter, as my proposal clearly states. However,
AS A FEATURE, I allow the brain to be written in the same superportable
language, and AS A FEATURE thereby allow the brain to move from host to host
. Both these FEATURES are just suger-on-the-cream features, but they are
indeed very useful ones!
> - Mitra
>
> =================================================================
> Mitra
> [email protected] voice: (415)826-2499 fax: (415)826-4423
> <http://earth.path.net/mitra>
>
> Always remember you're unique, just like everyone else.
-- Hakan "Zap" Andersson |http://www.lysator.liu.se/~zap | Q: 0x2b | ~0x2B Job: GCS Scandinavia | Fax: +46 16 96014 | A: 42[email protected] | Voice: +46 16 96460 | "Whirled Psas" ------------------------------------------------------------------------ Never underestimate the bandwidth of a speeding truck full of DAT tapes. ------------------------------------------------------------------------