From: broehl@sunee.waterloo.edu (Bernie Roehl)
Subject: Re: TECH: My standard is better than your standard.
Date: Fri, 10 Jul 1992 14:18:23 GMT
Message-ID: <Br6Fqo.FKu@watserv1.waterloo.edu>
Organization: University of Waterloo


In article <1992Jul10.022909.27996@u.washington.edu> lance@unix386.Convergent.COM (Lance Norskog) writes:
>Each deck will have to do its own collision detection for the "room"
>its user is currently in.

Do you mean collision detection between the objects "owned" by that deck,
or do you mean for *every object in the room* ?

>Each object will have its master copy 
>on some machine, and representative copies in each deck that happens
>to be in that room.

Sounds good.

>The representative copy can be read-only or
>read-write.  If read-write, it can be altered in the remote deck
>and a report sent back to the master.  The master can
>accept, reject, or consider several competing reports and
>average them.

Not sure what this buys us; shouldn't the "master" be the *only* one to
decide the state of an object?  We surely don't want multiple, divergent
copies of objects in different decks... everyone should share the same view
of the state of the world.

>If read-only, messages to the remote object do 
>nothing to it, but are forwarded to the master. 

I suspect all objects should be read-only.

>Consider modelling a game of fragile bumper-cars.  Remote bumper-cars
>are read-write in your deck, and when you hit them, they dent.
>They also tell their masters, "Ouch!  I've been hit and dented!".
>They don't break; that takes a change order from their masters.

I think denting also requires clearance from the master; otherwise you'll
see it dented in one way, and someone else sees it dented in another (or
not at all).  And whether it dents or not may depend on damage it received
a second or two earlier from a previous impact.

I think the "master" copy of each object should maintain all the state info
for that object; collisions should result in messages being sent to the
master copy, which decides how to respond and lets everyone know.  That way
all the decks have the same knowledge of the object's current state.

-- 
	Bernie Roehl, University of Waterloo Electrical Engineering Dept
	Mail: broehl@sunee.waterloo.edu OR broehl@sunee.UWaterloo.ca
	BangPath: uunet!watmath!sunee!broehl
	Voice:  (519) 885-1211 x 2607 [work]