Re: URL for interesting...

[glen e. p. ropella]

address@hidden writes:
 > I thought that Java was cross platform.
 > Is all of swarm going to be in Java, in the near future??? or just the
 > tcl/tk stuff?

Well, for now, it's just the GUI.  So, don't get too excited, yet! [grin]
But, there's still the possibility of moving to Java completely from
ObjC, when Java "comes of age", so to speak or when it's proven that
Java is capable of doing everything as well as ObjC.

Tony writes:
 > > While looking through the site, I initially dismissed the work their as a
 > > bit "toy-like," but realized that a direct comparison with Swarm did not do
 > > either project justice.  As I spent more time looking through the site and
 > > thinking about the research, it seemed that the ability to generate a Java
 > > applet that was a simulation *complemented* nicely what I have heard the
 > > Hive discuss (explicitness and publishing of experiments as code).  The
 > > learning, teaching, and web based aspects of the technology are intriguing.

You're quite right.  The concept of programming at a higher level
than C and programming behaviours instead of procedural/functional/
logical statements, is exactly what Swarm's end target is.  And this
is true, esp. if those behaviours can be dynamically constructed, 
mutated, and destroyed!

The one problematical issue is what I call "observability."  (Prepare
yourself for filisofickle nonsense.)  It is crucial to the extension
of simulation from a game-playing environment into a science that we
maintain complete observability of the system.  This means that if
I design a system that has 4 tiers of observability, say, a Swarm
containing Swarms and objects, containing Swarms and objects, containing
objects, then I should be able to take measurements on that Swarm during
it's forward iteration on any of those four tiers, the additional tier
of the event structure for the whole sim, the additional tier of the 
operations of the language upon which everything is based (the runtime),
and the state of the process upon which the runtime is implemented.  I
might even go so far as to say that the state of the machine upon which 
the processes are implemented should be observable.

With a Swarm application, the Swarm system, the ObjC language, and the
unix OS (esp. linux), I have more guarantees of this observability than
I do with, say, ProGraph, a visual programming language, where an
obfuscating translation occurs between the design of a system and the
execution of the system.  Even in such an excellent system like ProGraph,
where it *is* the visually programmed system that you see running on
the screen when you execute it (which makes it totally unlike packages
like Visual Basic or Visual C++, where the "visual" object that is created
is just a visual representation of a textual program), it still goes
through some semantic translation between what the user is executing
and what the machine is executing.  Of course, this problem would go 
away if there were a source distribution for ProGraph such that one
could muck around in that translation and take data at that tier...

So, while this AgentSheets project *definitely* looks like its in 
the right vein for Swarm's goal of establishing a common language
and platform so that model publishing is subsumed into modelers'
subconsciouses by catering to non-programmers, we want to be careful
not to shoot ourselves in the foot and make a "simulation science"
more difficult to achieve by severing the observability of the models'
execution.  And I have no idea how opaque the "agentsheets->Ristretto->
JavaVM->OS" process is.  We have Macs here at sfi, so, in my copious
free time, I'll check it out.

 > > One issue is that the compiler is only available for macs/powermacs (I
 > > think).
 > > It would be nice to hear what others think about this technology and some
 > > creative ideas about applications, "linkages" with Swarm, and the evolution
 > > of Java and Swarm.

Thanks for pointing this project out, Tony.  It's worth keeping an 
eye on.

glen
-- 
{glen e. p. ropella <address@hidden> |  Send lawyers, guns, and money!  }
{Hive Drone, SFI Swarm Project         |            Hail Eris!            }
{http://www.trail.com/~gepr/home.html  |               =><=               }


                  ==================================
   Swarm-Modelling is for discussion of Simulation and Modelling techniques
   esp. using Swarm.  For list administration needs (esp. [un]subscribing),
   please send a message to <address@hidden> with "help" in the
   body of the message.
                  ==================================