> I am not teaching this Summer
> specifically so that I can do the development work
> which I have been putting off all this past year-and-a-half
> due to the press of getting used to teaching...
I love the Summer too! No students, loads of bandwidth, prep not due till
September = Loads of time for the play that makes up for the stress of
teaching!
> I will be doing a major revision of VNet over this Summer
> which I am tentatively calling "VNet2"
Hurray!
> my major goals for this rev are:
> o persistent users and user-states via JDBC
There are more issues of security and memory handling here (e.g. if a client
gets cut off and reconnects, how long to keep the state info and how good is
the garbage collection with lots of orphan connections.
> o extreme scaleability improvements [perhaps to "massively-multiuser" level]
> o extreme efficiency improvements
I agree with another poster that some handshaking between multiple machines
is a good bet for this, but don't necessarily agree with another suggestion
of making the client "thicker". Generally, computation is cheaper than
bandwidth, but with 'lots' of users, local processing can only go so far.
> o one server managing multiple worlds
> o no-logoff/on for teleportation between worlds if managed by same server
This is the one feature of Deep Matrix that I quite liked (though there were
others I did not). Are you planning to have cloned worlds like the aforesaid
chess world that allows only two participants, while newcomers are directed
to a copy to enforce the max number of members.
