🚀 go-pugleaf

In article <DFI1rr.LyC@undergrad.math.uwaterloo.ca>, kominek@links.uwaterloo.ca () writes:
|> In article <DAGENAIS.95Sep22100113@gutrune.vlsi.polymtl.ca>, dagenais@gutrune.vlsi.polymtl.ca (Michel Dagenais) writes:
|> > 
|> > I finally got the exact reference from one of the people posting on this
|> > thread:
|> > 
|> > Laurie Griffith, "Modula-2 is three times less error prone than C",
|> > Proceedings of the Second International Modula-2 Conference,
|> > Loughborough University of Technology, UK, September 1991, pp 332-338.
|> 
|> 
|> The kicker is: why did not IBM (the source of this paper, I gather),
|> heed their own research? They should. The reason why IBM delayed 
|> introduction of the PowerPC personal computer is quite transparent.
|> Because they are unable to port OS/2 in a timely fashion. One 
|> programmer in the know called it a "big bag of dirt," i.e. a 
|> convoluted mess of C and assembly. Viewed from the outside the
|> operating system is wonderful. But on the inside ....

There are numerous reasons why IBM didn't adopt Modula-2.  My team leader
was involved in the "council" that critiqued various languages.  I was
(and still am) involved in Modula-2 compiler support so I got to provide
input.

A number of languages were looked at including Objective C, C++, Eiffel,
Ada, Modula-2, Modula-3, and a few more I can't remember now.  Some trial
projects were underway in a number of the languages and the outcomes of
these projects were used in the evaluation.  All the languages had
good and bad points but overall C++ looked the best.

Some things that killed Modula-2's chances were:
1) Not an OO language.
2) The glacial pace of the standard.
3) Difficultly in porting between systems because of compiler differences.
4) Lack of good compilers on needed targets.
5) Lack of skill in new hires.

The Modula-2 pilot projects we did locally went very well but the problems
(especially #1) pretty much killed any chances of further use.  Some of the
(now) old code is being enhanced by rewriting it in C++.'

Other things killed other languages chances.  Modula-3 looked really good
but fell flat on the available implementations and massive (at the time)
system requirements of the compilers.  At the '91 (I think) OOPSLA meeting
I talked with the Modula-3 SIG group about this and they didn't understand
why everyone didn't just buy a top notch workstation to do development.
They didn't see the real need for a Mac or PC implementation.  Oh well,
they missed the boat and it's too late now.
-- 

- Bill Seurer     ID Tools and Compiler Development      IBM Rochester, MN
  Business: BillSeurer@vnet.ibm.com               Home: BillSeurer@aol.com