Re: getting input from human-agents via swarm/tkObjc

[cgl]

Hi Rick!

>  What I'd like to be able to do is have some of
> these agents choices be input by users sitting at
> different displays, so they can interact in the world
> with the other computer run agents, but not see
> what the other humans are doing.

This is a specific use of a more general capability that we 
obviously want to have in swarm: the ability to create
swarms on multiple processors over a network that communicate
with one another. 

 We have had quite a number of discussions on how to 
do this, but have not had the time to delve into 
implementing any of them yet. 

  The general consensus is to incorporate the "Distributed
Objects" and "Remote Messaging" capabilities described in
Next's excellent manual on objective-c. To get an idea
of what this is all about, you should take a look at
those sections in their online manual at:

http://www.next.com/Pubs/Documents/OPENSTEP/ObjectiveC/Objctoc.htm

When we finally get to the parallel version of Swarm, we will 
probably be implementing it something along these general lines.

Another possibility is to embed the functionality of a distributed
message passing scheme like MPI.

  If y'all want to take a stab at it, or if anybody else wants
to, let's try to coordinate our thinking and design goals. Clearly,
this relates to the capability to invoke multiple versions of
the same model, ala Drone, but *within* Swarm, as we have been
trying to get to for some time with "Experiment" swarms, which will
manage multiple, independent, invocations of the same model.
Distributed swarms would be more interactive with each other, and
could be different from one another as well.

Anyway, check out the sections of the Objective-C manual referred
to above and post your reactions..

Cheers!

Chris Langton