Article 172650 of comp.os.vms:
In article <33A78C85.3B4@cri.dk>, Arne Vajhoej <ava@cri.dk> writes:
>
>I agree that C++ programs written for Unix or for MS-Windows is often
>absolutely non-portable.
>

Sadly, that covers about 99% of all programs.

>But that is not C++'s fault. There are clear standards. People can
>choose
>to follow those standards and then their programs will be relative
>portable
>or they can choose to ignore the standards and the result is the usual
>mess.

It's not so simple, the "standard" is not so fixed as one might like.  I'm
working with cross platform C++ GUI called Amulet. The Mac version would
build with Metrowerks 10, but so far, no other version since then.  Reason:
Metrowerks changed the "standard" libraries.  Apparently their new
implementation of "bool" is incompatible with that in VC++ and DEC C++, 
along with assorted and sundry other problems. My understanding of this is
that Metrowerks had been shipping one "branch" of the standard library
specification, but now have moved to the other "branch", to bring it closer
to other compilers (or the one true spec?) Anyway, at the moment, there is
no way to build Amulet on Mac (or Windows) with the current Metrowerks
compiler. There is a fellow in the Amulet project working on it, and I'm
pretty sure the resolution will be recoding sections of Amulet to allow it
to work with the compiler.  This is NOT portability! 

Amulet is the ONLY C++ package I've ever seen that ever ran on Mac, Windows,
Unix, and OpenVMS.  So it is true that people can write portable C++ code,
but portability does not seem to be a natural consequence of using that
language, as it is for, say, ANSI C.  In fact, it appears that the only way
to have reasonable confidence that your C++ code is really portable is to
actually  build it on different platforms. 

From another post by Blair Phillips:

>Have a look at "Comparing Development Costs of C and Ada", written by
>Stephen F. Zeigler, Ph.D., of Rational Software Corporation.
>Available on the web at:
>http://www.rational.com/support/techpapers/c_and_ada.html

That is a fascinating read!  In this case study, C is shown to be about 50%
more expensive to code in than Ada, and is much more bug ridden.  Anybody
suprised? As for C++: 

  "Bug rates in C++ are running even higher than C, although we have no where 
  near the ideal comparison platform that we have had with VADS for C and Ada."

It simply amazes me that while companies spend many billions of dollars on
software development, there is apparently exactly one careful study of the
relative costs of the different languages.  I mean, in most fields of
engineering the cost of the processes are important, and minimizing those
costs is a big part of the effort. Apparently in software development using
the currently popular and "cool" language is more important.  This just
confirms my suspicion that "software engineering" is a misnomer, and that
software production is still basically an art form.  Besides, who cares if
the Emperor has no clothes, so long as they are the latest and trendiest, eh?
(Certain operating systems come to mind.)

Regards,

David Mathog
mathog@seqaxp.bio.caltech.edu
Manager, sequence analysis facility, biology division, Caltech 
**************************************************************************
*Affordable VMS? See:  http://seqaxp.bio.caltech.edu:8000/www/pcvms.html *
**************************************************************************