[comp.lang.smalltalk] Discrete-event simulation in Smalltalk-80

cameron () (05/08/90)

Hello, netters.

Has anyone out there done any DES in Smalltalk-80? We're just starting
on a simulation project, and I'd like to find out how others have
approached this area. Specifically:

	1) Did you use the ParcPlace classes (Simulation, SimulationObject,
	   etc.) as a basis? Did you add new classes/new methods? Are you
	   willing to discuss/share them?
	   
	2) Do you know of any books/articles which discuss DES in ST?
	   I've read what the purple (aka blue) book has to say, and
	   I've found a couple of introductory papers in OOPSLA pro-
	   ceedings, but otherwise nothing.
	   
I'd be interested in hearing any comments or tales - including why
you chose to do OO simulation on a different platform.

Please reply by e-mail. If there's any interest, I'll post the
obligatory summary.

Thanks,
  Rick Cameron
  Mobile Data International
  Vancouver, BC
  ...!uunet!van-bc!mdivax1!cameron
  (604) 277-1511

madhu@hubcap.clemson.edu (Max) (05/09/90)

> Please reply by e-mail. If there's any interest, I'll post the
> obligatory summary.

*******Tried mailing it without success, so am posting this******

	  My Masters Thesis is a discrete event simulation prototyping
environment dealing specifically with :" Modelling Transient conditions
in Communication Networks" (The final version is in Smalltalk-80 ParcPlace)
and runs currently on the Mac IIx. Since standard stuff has been used (
no user prims) it should work in all environments supported by ParcPlace for
ST-80. 
 


	You can write to me at: madhu @hubcap.clemson.edu
				or\
				!gatech!hubcap!madhu
If you have any specific questions regarding the setting up of your
simulation environment, I would be glad to answer.

> Has anyone out there done any DES in Smalltalk-80? We're just starting
> on a simulation project, and I'd like to find out how others have
> approached this area. Specifically:
> 
	Yes, I extensively used ParcPlace's excellent Simulation class. I am
still mind boggled that the Xerox team could have thought of something so
complete and beautiful. You can end up recreating RESQ or PAW in a couple 
of days! 

I am not at home right now, but I remember that the Smalltalk
book by Pinston and Weiner(?) (The Tektronix version of Smalltalk) has some
elementary examples. Like in the blue book*, I would advice you to pick up
Birthwistle and just start doing what you want. 
*There is now a purple paperback version I guess.

	The elements you would give yourself would depend a 
lot on your end-application. If you want to create a generic resource-
contention-consumption environment suggest you get hold of a RESQ manual
and just recreate the RESQ elements. 

For Example if you are interested in Communication Scenarios:

You might wat to add a "destruction" event that getd rid of unwanted jobs.
in Communication environments this could save you a lot. Again, I am just
shooting off... what you need to decide are the tradeoffs and constraints
your simulation environment can have.

> I'd be interested in hearing any comments or tales - including why
> you chose to do OO simulation on a different platform.
Initially started out with Pascal on a VAX (VMS) system moved to C on
an Ultrix-32 Vax system, then since my job required extensive use of
Smalltalk thought what the heck ... give it a try. I now no longer use the
the other language versions. The OO part of it is the clincher since I
can do some really neat things that would otherwise require doubling my code
in a conventional language. There is the additional advantage using ST-80 
since it is a complete environment, reaql time data can be used, actual
network monitoring can be done...
> 
> Thanks,
>   Rick Cameron
>   Mobile Data International
>   Vancouver, BC
	Welcome
		madhu Raman
		Clemson University, Clemson, SC
		(Thesis under Dr. J.D. Spragins).
.