This is actually not as hypothetical as it soiunds. I'm working with a
firm that manufactures parts for nuclear reactors. Parts certification
alone can take years. I remember a friend working for GE in the 80's
remarking that their computers in the field were still built of discrete
semiconductors, with core and running from paper tape.
One has to keep in mind that companies are bought and sold, sell off and
scrap assets, assets are destroyed, etc. So keeping some sort of hardware
arouind to run this stuff is not an option.
So, here's my reasoning. What was the programming atmosphere like in 1975?
Well, punched cards and magnetic tapes were still very much in use (surely
everyone remembers Donald Knuth's "Searching and Sorting" text with the
foldout examples of how a polyphase sort worked).
COBOL and PL/I were very hot as was FORTRAN. There was a little bit of
renewed interest in the Iverson language, APL. Folks still programmed in
Algol, particularly in Europe. BASIC was being used in some teaching
environments, but wasn't really accepted as a full-fledged production
language. LISP was popular among the research types. Wirth was flogging
Pascal, but it still looked like something for the academic community
(i.e., there wasn't much in the way of I/O). There were various
special-purpose languages wandering around, like SNOBOL and Simula (not to
mention GPSS). RPG was far from dead. I was still poring over the
occasional bit of JOVIAL. But if I wanted to do something "quick and
dirty", it was in FORTRAN. There were whole operating systems written in
extended dialects of FORTRAN. And there were lots of "boutique languages".
All this time FORTRAN has survived. Still a great language for number
crunching, it's withstood attempts by X3J3 to make it a completely opaque
mess of features. By and large, if you write a program in FORTRAN IV or
77, you can find a compiler for just about every machine ever made.
One nice aspect of FORTRAN is that it's easy to learn a functional subset.
No trying to understand multiple inheritance and objects. Syntax is very
simple.
This was driven home to me when I picked up an 8" diskette from my
collection written in 1978. I'd written a program for CP/M that read and
wrote ISIS-II floppies. Aside from the basic physical diskette I/O, it was
entirely in FORTRAN. It was easy to understand even though I'd not written
any FORTRAN in 20 years.
Cheers,
Chuck