Re: internal-vs-external behaviors
Peter J. McCann ([email protected])
Mon, 24 Apr 1995 22:25:46 -0500 (BLT)
Gavin Bell writes:
> On Apr 24, 3:26pm, Peter J. McCann wrote:
> > An external behavior mechanism is absolutely needed for multi-user
> > interaction of any sort.
>
> Nuh-uh!
>
> Multi-user interaction might be as simple as:
>
> SharedObject {
> sharedObjectRendezvous "server.sgi.com:World15port"
>
> objectToBeShared USE GavinAvatar
> }
>
> ... where SharedObject is just an object that takes care of all of the messy
> protocol stuff needed to do multi-user interaction, and GavinAvatar is just
> an object with behavior (maybe my representation in the virtual world...)
> that should be shared between worlds and is DEF'ed somewhere previously in
> the scene.
>
> All inside the scene. You do need a protocol, but we already have protocol
> INSIDE vrml.
>
> I'm not saying this is the best way of doing it, I'm just throwing this out
> to contradict your statement that multi-user stuff has to be done outside the
> VRML object model.
It's true that some kind of policy has been defined by this construct,
but no mechanism for doing so has been discussed. Whatever mechanism
this is, it could be called an "external behavior." Although I agree
it would probably not be part of a VRML spec (in this case), it would
be a part of a VRTP spec, and I think each should be designed with the
other in mind. I think encapsulating all the "messy protocol stuff"
as someone else's problem is not reasonable, even for a standard which
doesn't immediately address that stuff.
"External behaviors", under any implementation I could envision, would
at least have to define some kind of interface (maybe with CORBA IDL)
that crosses the browser boundary. I think this interface would be the
same one used for interaction among users, at least at some level. It
might even be the transport mechanism itself.
-pete
--
Pete McCann [email protected]
Department of Computer Science http://swarm.wustl.edu/~mccap/
Washington University in St. Louis