[ODE] Re: ODE on multiplayer games

Henrik Grimm hengr at hotpop.com
Wed Mar 23 19:13:41 MST 2005


Megan Fox wrote:
> Either way of approaching it (either all server-run or with the owners
> spread througout the network), how do you manage to avoid
> inter-penetration in the process of gradual updates?
> 
> It seems as if you either need to update -everything- at once, causing
> a freeze, or you need to update gradually

I don't understand the difference between these.

> - but if you update
> gradually, what's to stop the case where you update box A (but not box
> B), where box A's updated position interpenetrates with box B's local
> position, causing a local physics explosion, then you update box B,
> but box A is already flying away at high speeds, causing more such
> explosions... etc?

All nodes have their own physical simulation of the world, and these may 
deviate somewhat. If a node gets a state update, it does not need to 
replace its surrogates state, but can instead try to make the 
surrogate's state approach the extrapolation of the state update 
(perhaps by adding a force).

In other words, you have multiple replicas of the simulation state that 
diverge with time, and state updates are used to make them converge.

/Henrik



More information about the ODE mailing list