Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!sdd.hp.com!decwrl!bacchus.pa.dec.com!decvax.dec.com!zinn!nuucp
From: mjv@objects.mv.com (Michael J. Vilot)
Newsgroups: comp.lang.c++
Subject: Re: Advice on design
Message-ID: <908@zinn.MV.COM>
Date: 2 Sep 90 19:59:09 GMT
Sender: nuucp@zinn.MV.COM
Lines: 40

I said:
> Eventually, I think that we will develop an approach to design that is
> uniquely suited to C++.

Ralph Johnson replied:
> If true, this would be very bad.  Design is supposed to be different
> from implementation, so it would be odd if every language required a
> unique design approach.

I think this is an interesting discussion, and I certainly don't claim to
have any final answers.  I've observed that the design approach influences
the way we use a language, and the facilities of the language in turn
affect how we formulate designs.  I don't think it's the case that the
language ``requires'' a certain approach, it just seems to evolve one.  Over
time, different language cultures evolve idioms and conventions that seem to
work well enough that they become widely used.

It's certainly possible to come up with a very abstract, language-independent
design.  Would it make the most effective use of the language at hand?  
I think not -- this seems to be the same portability/efficiency trade we
encounter at the coding level.  On the other hand, this implies that the 
same sort of Pareto analysis holds: the majority of the design could be
language-independent.

> Booch's book is good.

I think the book is a landmark work, because it is the first one to illustrate
the object-oriented design approach in a way that transcends the features of
any one language.  However, I see that although it's possible to use OOD to
design into different languages, I do not see that the _same_ design is 
expressed the _same_ way for every language.  Some research (and a controlled
experiment or two) would help provide some substance to the discussion.

Perhaps having a method like Grady's OOD will short-circuit the usual formation
of informal idioms that we've normally used as ``design method.''  It will be
interesting to see if his approach catches on.

--
Mike Vilot,  ObjectWare Inc, Nashua NH
mjv@objects.mv.com  (UUCP:  ...!decvax!zinn!objects!mjv)