> Here are just a bunch of thoughts I have been having on the subject of VNet
> possibly becoming distributed, or peer-to-peer, VR. That is, not having any
> one central server -- using a distributed system of interconnected servers
> like the web does.
not technically "peer to peer"
but IMHO def the right idea
[and somewhat where have been headed
with the next-alpha rewrite]
thanks miriam
> Imagine a VNet universe where many worlds join up into a larger single
> world. You could walk seamlessly from one world into another -- they would
> look like one large world. A distributed VNet universe would live on many
> machines at once, just like the web. Different parts would live on the web
> pages of the various owners of worlds. The owners would choose what other
> worlds they wanted their worlds to link up with, just as web pages link to
> other web pages as their owners wish. (Of course there would be no
> requirement to link worlds together if their owners wanted them to be
> separate, but the facility would be there if desired.)
the direction I have been thinking in
is having the distributed or integrated servers
share a database backend...
that way they can share
username/password
default avatar
the idea I have been mulling over
is to add a "teleport" node/PROTO
which causes the server you are currently logged into
[call it "foo"]
to [try to] pass control over to the new server
[call it "bar"]
given that we continue using an applet
[java in a web browser conext]
and given the java security model
[an applet can only open a socket
back to the machine from which it was served]
this would mean that
a) foo would need to act as a pass-through
to get/put messages between the client and bar
[foo becomes a middleware layer
between the client and bar]
b) contrary to the current setup
[currently the server knows nothing
about the scene being used]
any given server would need to
have a sense of what wrl [URL] it "owns"
so it can pass the replacement scene along to
the new client through the middleware/mediator server
> Each world would have a constantly updated list of addresses of the current
> visitors on its web page. But what is a user's address if, their IP number
> is dynamically allocated? It is different each time they log on. There are
> a few ways we can get around this problem.
see above discussion of VNet2 acting as a middleware layer
-----
I am in the middle of
end-of-the-quarter grading craziness
but
"the plan" is to finish up the next server-alpha re-write
over the quarter break
finish adding a DB backend
and thus persistent username/password
and persistent default-avatar by username
and last-known-location by username
[location at next logon == location at last logoff]
and message-object pool [to increase server speed]
the last server-alpha just added:
upgrade server code to JDK 1.1.x
add code packaging
and some code cleanup...
it is currently running at
http://mercury.it.rit.edu/~jeffs/vnet/
and has been "banged on" some
thoughts/reactions??
jeffs
--
Jeff Sonstein
Assistant Professor of Information Technology
Rochester Institute of Technology
---------------------------------------------
http://ariadne.iz.net/
http://www.it.rit.edu/~jxs/
http://ariadne.iz.net/~jeffs/jeffs.asc
=============================================
there are no bugs
there are just undocumented features
