I'm not sure what you're asking or asserting, Paul.
Yes, the Scobleizer blog was good.
Re P2P: Ummm, define P2P. I don't know how to do that. I can tell
you how Croquet works:
Each participating peer completely computes their own results. The
Croquet model keeps the simulations in sync based on the idea that
the same inputs at the same time will produce the same results. The
current version of Croquet uses a technique called "Simplified Tea
Time" to provide the notion of "inputs at the same 'time'."
Specifically, each participating peer sends its inputs (e.g., mouse
movement, keyboard press), to a "router" which timestamps the input
and distributes it to each participant.
Is this P2P? I don't know. If it isn't, then I guess nothing on the
Internet is P2P, because the Internet uses routers to transmit
messages. So does Croquet. Just like IP hardware routers, Croquet
routers do no computation: they just timestamp messages and forward
them.
Croquet routers are implemented in software. Any machine is fine,
including one of the participants.
For practical purposes, it is convenient to have the router be on a
fixed IP address relative to each participating peer. (Experimental
version of Croquet have used hole-punching techniques that did not
require this, but I don't know of anyone working with this approach
right now.)
The Croquet SDK code broadcasts router locations on the LAN. Over a
Wide Area Network, various other techniques can be used, including
just hardcoding a fixed router address. The SDK also has tools to
display your router information as XML, which can be sent via any
external means such as email, text chat, etc.
WiscWorlds and a related current project, KidsFirst (http://
www.wetmachine.com/item/684), do make use of what we call "continuity
servers." These are nothing more than a router on a fixed IP
address plus an ordinary participant that stays connected forever,
and thus provides continuity even when the last person leaves and
comes back later. Maybe we shouldn't call it a "server"?
Eventually, I could image distributed routers -- what might be called
"Full Tea time" -- in which each peer acts as node in an overlay
distribution network. The message delivery aspects of this are fairly
straightforward and covered in academic literature. But frankly I
don't know how to do the time coordination part. That's a question
for David Reed...
Howard Stearns
University of Wisconsin-Madison, Division of Information Technology
On Jan 8, 2007, at 7:28 PM, Paul Sheldon wrote:
> There was some misleading stuff, I believe, about croquet locations
> being meaningful over internet without a server of a Wiscworlds,
> but allegations are to the contrary indicating peer to peer
> collaboration with no need of server :
>
>
http://scobleizer.com/2006/05/06/wow-3d-operating-system-open-croquet/>
> Action item :
> IT people at Redlands and Pasadena library finding time with me to
> fool with this concept. I've only done multiple installs at Apple
> Willowbend Store and had wifi P2P, not internet P2P, yet.
>
> Have high speed access person away from IT control worries.
>
> Brains fried. Composition so far bad. Sorry.
Locked
