Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!seismo!husc6!cmcl2!beta!hwe
From: hwe@beta.UUCP (Skip Egdorf)
Newsgroups: comp.lang.smalltalk,comp.lang.c++,comp.lang.lisp
Subject: Re: language issues (was Re: teaching object-oriented simulation & modelling)
Message-ID: <6440@beta.UUCP>
Date: Wed, 17-Jun-87 12:43:59 EDT
Article-I.D.: beta.6440
Posted: Wed Jun 17 12:43:59 1987
Date-Received: Sun, 21-Jun-87 17:01:47 EDT
References: <7116@linus.UUCP> <1995@sphinx.uchicago.edu>
Organization: Los Alamos Natl Lab, Los Alamos, N.M.
Lines: 49
Keywords: availability, speed, portability, graphics
Xref: mnetor comp.lang.smalltalk:162 comp.lang.c++:337 comp.lang.lisp:291

Summary:We do Simulation in an O-O environment.


In article <1995@sphinx.uchicago.edu>, code@sphinx.uchicago.edu (paul robinson wilson) writes:
> In article <7116@linus.UUCP> sdl@linus.UUCP (Steven D. Litvintchouk) writes:
> >
> >We may soon have a need to train several engineers in using
> >object-oriented techniques for discrete-event simulation and
> >modelling.  (Languages such as Smalltalk-80 and Simula-67 come to
> >mind.)
> >
> >We are therefore interested in finding out if there are any courses,
> >tutorials, textbooks, etc., that are good introductions to this
> >subject.  (Note what we are looking for is how to do object-oriented
> >simulation and modelling in Smalltalk or Simula as opposed to software
> >development.  We know of lots of books that teach simulation in
> >Simscript and GPSS, but not simulation in Smalltalk or Simula; other
> >than the Blue Book, which barely touches on the subject.)
> 
> I too am interested in this, and have several other questions as well:

My group has been doing large discrete-event simulations in an Object-Oriented
environment for the last couple of years.

The main overhead is that it takes a good simulations person 6-9 months to be
trained to use these environments and methodologies. We have a Software
Engineering methodology based on Actors that is used for in-house training
that we developed precisely because we could find no information on how to
develop simulations in this environment.

We use Intellecorp's KEE on Symbolics, TI explorer, and (Lately) Sun
Workstations.
We have found KEE to be an excellent base for such work, with productivity
several times that of more conventional environments.
The KEE environment, while very rich, is not any sort of standard. I do not
believe that a standard exists that encompasses the necessary tools, graphics,
etc. required.

I think that the field might be ripe for a how-to book on simulation in the
Object Oriented environment. The biggest problem is just what environment
to describe??
Smalltalk is the closest thing to standard that includes graphics.
However, one of the biggest advantages that we have found to the KEE environment
is that an "AI" base that includes forward chaining allows us to attack
problems that include simulations of the human decision making process.
These simulations are just too hard to do in more "classic" environments.
Do we write a Smalltalk forward chainer for inclusion in such a book??
Or is there another environment out there that includes O-O, graphics,
a super development environment, and "AI" things??

					Skip Egdorf
					hwe@lanl.gov