HTTP should be able to handle generic VRML scenes. However, if you want
to create an entire virtual world using VRML you really need to implement
your own server. What we should really do is create a standard set of
interactions with a VRML world server, so that no matter the driver
implementing the world (MUSH/MOO/ColdX/etc), the client can use it.
< I'm assuming the VRML server will work like HTTP, where a browser
< connects to it once, downloads all the pertinent data about a VRML room,
< then disconnects from the server. If this is so, then I assume it won't
< be possible for multi-user worlds where you can see other people moving
< around?
In this stage, my personal answer (off my interpretations of the language
and what is said on this list), is "maybe". The implementation I am
creating (and I believe I can finish in a month, with help on the client
end) has "stream" interaction (aka something updated constantly) through
a textual interface and static interaction through VRML (where the scene
may change, but you will not know unless you reload it).
However, I do believe that VRML 2.0 will correct these problems (although
I dont know how, it seems like VRML would need to define an update
feature, where you can update an existing object in an existing scene,
then define a server protocol to send the update through a stream).
As for your other questions, I'd suggest reading the VRML 1.0 specification
at:
-Brandon Gillespie-