Re: Swarm Design

[glen e. p. ropella]

> As I start to get a better idea of Swarm, one question that continually
> comes up is, "Why was objective C chosen for the implementation of Swarm?"
> I would have expected a more mainstream language such as C++ to have
> merits. Mostly I'm just curious about the rationale of the developers.
> 
> My other question is "Does Swarm have any objects for genetic or NK
> models?" Is anyone working in these areas with Swarm?
> 
> Many Thanks.
> 
> Paul Black

Paul,
Can I suggest that you do a search on our mailing list archive?
There's been lots of discussion about the advantages Obj-C has
over C++ (and some more recent discussion about Obj-C vs. Java).

And this will be in the FAQ, once I get some time.  But, basically,
the capability Swarm needs that is either lacking or difficult
in other OO languages is the dynamic construction and dispatch
of messages (or function calls, for us simpler folk).  This 
underlies Swarm's mutable schedules, which blossom only while
a model is running.  C++ forces the programmer to define all
the types of messages that are to be sent at compile time (via
virtual functions, I believe).

Now, Java supposedly has a new API that allows one to do what 
Roger has done in the activity library.  But, I, personally,
haven't read it.  See

http://www.javasoft.com:80/products/JDK/1.1/docs/relnotes/features.html

particularly the Reflection API.

As to your second question...I'm pretty sure the answer is no.  And
I'm not really sure that Swarm is the best platform for doing that
kind of modelling.  But, I don't know enough about it to say, really.

glen