Re: [Swarm-Modelling] Something Glen said

[Matt Murphy]

One quick thought on this, as someone who had a fairly tough time
understanding enough about Swarm (see the mailing list archives) to
debug extremely simple modeling issues:

While Javascript is a great language, I think a language like Ruby
would be even simpler for those new to programming.  Ruby has the
following nice features:

- syntax is extremely simple and concise
- great for writing domain specific languages (it's easy to write
syntax that isn't really part of the language but feels like it).
- runs on both the .net CLR and the JVM (projects currently in beta)
- extremely dynamic, makes it easy to add new behavior to individual
instances, whole classes, etc., at runtime.
- has a distributed ruby extension that would allow some types of
parallelism to work very well on clusters.
- supports inline c and c++ for performance critical stuff.

Ruby is slow compared to compiled languages and uses way more memory
than C#, Java, etc., but I'd argue that most modelers don't really
need tons of performance until after they've fine-tuned the model's
behavior, a phase that would go a lot faster in a friendly, dynamic
language.

I think for people who are new to modeling or who just want a model
whose source code is extremely self-evident (perhaps nearly readable
to the layperson), Ruby would be a great choice.  Sounds like it's
also compatible with the Mozilla XP-Com stuff Marcus mentioned.

(this has been a great discussion, just thought I'd throw in some
thoughts based on my recollections from when I was learning Swarm and OOP.)

-Matt Murphy

On 11/27/06, Marcus G. Daniels <address@hidden> wrote:
> glen e. p. ropella wrote:
> > And I also agree that Objective-C (and Swarm's
> > idioms but not the current software) is a better foundation for future
> > ABM work than Java or .NET.
> >
> I don't agree with this.  When I say this,  I am not saying that the
> path to a glorious future ought to involve the current Swarm software.
> I'm saying that if Objective C is desired, and considered necessary for
> sufficient flexibility in modeling, then we might as well update,
> refactor, extend the current software as it is still in maintainable
> shape.  Objective C is now back from the dead thanks to Apple, so we
> could contemplate making changes to Swarm to use its runtime instead of
> the embedded one.
>
> This talk of object evolution, to the extent it is actually crucial to
> facilitate modeler productivity, is not hinged on use of a dynamic
> language like Objective C.  It's common in many programming environments
> to have reflection (the ability to programmatically look over the stuff
> in a running program), and that's all that's really needed to have
> dynamism (as is done in Swarm for Java).
>
> I'd say if we want to break free, we should do take some time to study
> modeling and computer science literature.  Objective C is a fine
> language, but hardly the state of the art.
>
> Also, I think it is a mistake to ignore the ways the .NET CLR and Java
> JVM may advance in near future.  C#, for example, is adding advanced
> features a rapid pace and the JVM Hotspot virtual machine is very highly
> studied engineerring, and software that is now GPLed.  All kinds of
> innovations are possible, the Swarm community could even contribute to them.
> _______________________________________________
> Modelling mailing list
> address@hidden
> http://www.swarm.org/mailman/listinfo/modelling