Re: Swarm technology preview

[Marcus G. Daniels]

>>>>> "RP" == Rod Price <address@hidden> writes:

RP> I did want to comment, however, on the distinction you seem to be
RP> making between static, strongly typed imperative languages and
RP> dynamic, weakly typed declarative languages. 

I'm using "dynamic" as measure of the degree to which components are
forcibly, globally checked for type consistency.  My experience is
that it can be useful for exploration to keep things loose, at least
for a while, like not pounding in nails all the way.  I'm not
inclined to think the real world is a perfect system of types, either.

RP> Personally, I think that this combination of strong typing and 
RP> declarative programming is very powerful.  I find that if I can
RP> get a function to compile, it will usually do what I expect.

For other kinds of programming (maybe even most) when you have a very
clear goal in mind, strong typing is very welcome.  The declarative
aspect makes it all the better.

I'm not really facile in Haskell, ML, Mercury, etc., but it does seem
clear to me this would be a solid way to express many kinds of models.
It would be great to work with an expert in functional programming in
order to get something *good* working (OTOH, I could probably get
*something* working).  I don't think it is especially important to
co-evolve the language infrastructure with the high level semantics.
I think figuring out well-defined-vocabularies can be facilitated with
an XML/DTD like approach, independently.

RP> There are several
RP> declarative (actually, functional) languages which are strongly
RP> typed, yet dynamic in the sense that they use higher-order
RP> functions. 

Hmm, even C functions can take functions as arguments.  As a practical
matter, higher order functions in Lisp are easier to use and more
flexible than higher order functions in C because introducing new
functions is so lightweight.


                  ==================================
   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.
                  ==================================