C++ as "the revision of C"

[uunet!labrea.stanford.edu!cygint!gnu]

Recent comments encourage me to temporarily abandon my lurker's role on
numeric-interest.

C++ is not a revision of C.

I have been making some major structural changes to a program lately
(gdb), adding transfer vectors where direct calls used to exist.  It
was pointed out to me that I could have converted to C++ and made a
class instead.  After a moment's reflection, the problem is:  C++ won't
compile my 99,000 line program without major changes, since the
program is written in C.  Not ANSI C.  C.

My view on the whole lemming-like trend toward C++ speaks for itself.
The language defiles the concepts that underlie C, like simplicity,
generality, and predictability.  The parts of C that were poor and
complex were retained, like the declaration syntax; and the parts that
were simple and strong, like the name spaces, function calls,
conversion rules, and parsing, were mangled and complicated.

I encourage NCEG to handle numerics in C rather than abdicate and run
off to C++.  A parallel NC++EG might want to look at numerics issues
there, though I doubt there are more than five people experienced
enough to have something useful to say about numerics issues in C++
programming.  Still, it'd be hard for me to fault numerics design by
committee in C++, since the existing design by fiat has not produced a
language to my taste either.

	John Gilmore