Re: [www-vrml] some questions about X3D

[chris <chris@surak.com.au>]

Paul Aslin wrote:

> Wouldn't it make more sense to provide a ECMA script
> Browser call which would send data packets to the remote
> server ?
>
> Say:
> Brower.sendDataPacket(data,server,port,method,loginID);
> where method is UDP etc. presumably we can use use
> eventsProcessed or Brower.addRoute... to receive
> infomation.
>
> See that way developers would be free to implement whatever
> systems they needed, as opposed to having to try and alter
> an existing system which is coded in the X3d brower itself.
>  
If it is an SAI Ecmascript call the browser would still have to 
implement it, but I think your
idea is a good one.  sending a DIS PDU could have been implemented on 
such a thing
and it gives the option of using the protocol that best suits the 
application needs.

My first thoughts were along these lines and it should be explored more, 
I think.

More on the web3D browser implementation: it would have to be assured 
that the call could leaverage the
existing network http infrastructure.  That is, if http is the method 
then you would want URL encoding with
parameters such that a JSP, servlet or CGI can use existing mechanisms 
to pull the parameters off the request
in the normal way.  So, for http you would want "parameters" instead of 
"data" and the parameters would be
an MFString.

> One further advantage would be to allow connections to
> multiple server's at once.
>  
yes

> I have tried doing something simple with Blaxxun Connunity
> server, simply trying to change the LOD distance used to
> SharedObjects has proven impossible because the code to
> write the objects is part VRML and part OCX plugin.
>  
I'm sticking to generic methods at the moment to avoid this sort of thing.
I'm expecting to run up against performance problems for some interactions.
My understanding is the VOS people dumped http communication because it 
was too slow.
DIS uses multicast for its efficiency (over something like http) and 
because it suits the application.
Considering these facts I think there may come a time where we can see 
X3D does not
support generic network communications that are efficient enough for 
many applications.
That remains to be seen - I'll keep soldiering on with the tools 
available and see what falls out at the other end.

> Should we stipulate/provide a way of allowing X3d scenes to
> intercept network communications, rather than having to try
> to change things later on.
>
> By this I basically mean that multi-user addon system
> should add items to scenes via a node rather than directly
> which creates problems with syncronzation of data between
> what is in the scene and what is in Script nodes.
>  
I don't follow.  You seem to be going back to the Node style proposal of 
Miriam's here
as apposed to a SAI call.

regards,

chris

> -
<snip>

-------------------------------------------------------------------------
for list subscription/unscrubscription,
go to http://www.web3d.org/cgi-bin/public_list_signup/lwgate/listsavail.html