Re: Behaviours (Was: Re: ADMIN: VRML + JAVA - A Wedding)

James Waldrop ([email protected])
Wed, 18 Oct 1995 08:17:20 -0700


>We all agree on the "single point of control". All "objects" in the scene
>must have exactly one brain, or control point.

Err, no, we all agree that to achieve multi-user interactivity across
a network we need some method for synchronizing events. This is a
technical requirement. Your statement is in fact what I would consider
a bad design. A good design allows the network to degrade gracefully.
If you use the canonical windmill example, the windmill hitting someone
does not have to generate traffic in some designs (the vector position
of the person it hit is another matter). One is deterministic and
by its nature re-entrant. The other is nondeterministic in the extreme.
By degrade gracefully, I mean that if I miss a few packets that show
you walking into the windmill for 500+ms, I don't care too much, because
in my world you haven't walked into it for it to hit you. If this
windmill is a rsally key feature of the landscape, then the collision
can be "synchronized" with all its copies (or however that works, all
the good proposals have a method for this).

A good design will make sure that mostly non-deterministic messages are
sent after the world is loaded.

James

--
James Waldrop				/   Technical Director
[email protected]		/	    Construct Internet Design
[email protected]		/		    http://www.construct.net

  • Next message: Mitra: "Re: DEF: invitation"
  • Previous message: Johnson, Jeff: "Re: VRML / QuickTime VR marriage"
  • In reply to: Master Zap: "Behaviours (Was: Re: ADMIN: VRML + JAVA - A Wedding)"
  • Next in thesad: Master Zap: "Re: Behaviours (Was: Re: ADMIN: VRML + JAVA - A Wedding)"