Re:Q: Why Are Scientists Still Using FORTRAN in 20 (Score 2)

I was really pleased that InfoJunkie777 raised this issue and followed with great interest the ensuing discussion. I suspect that not everyone who contributed comments has actually had much first hand experience with FORTRAN or even Fortran and some therefore are thinking out of the box. That's not a bad thing of course, since new ideas frequently spring from thinking outside the box. However, I would like to offer a few comments from someone who has been firmly inside the box. Take the issue of fast computing. When I started programming in FORTRAN IV back in 1967 processing speed was the least relevant issue because we were required to take part in a procedural loop from which, occasionally, there was no escape. This consisted of punching cards, submit the cards to an operator who ran the program in batch mode, an hour, or even a day, later collect a printout, correct punching and programming errors, resubmit, and keep going around the loop until, with luck, finally escape with real results. Computing speed in this regime was effectively instantaneous. About the same time some bright spark (who would hate to be named now) invented a version of FORTRAN called PORTRAN (Poor man's Fortran) which was used by kids in rural schools with no direct access to a computer. This involved programming by using a paper clip to remove chads in blank punched cards and posting them to a computer operator in a city. I know it sounds dreadful now, but this was a wonderful thing in its time, inspiring many young minds to go further in their education. I notice that many of the contributors to the discussion remarked that Fortran is still considered useful for scientists and mathematical manipulations, but nobody seemed to understand quite why. I think the reason is simply that it remains fit for purpose. All the criticisms of Fortran are perfectly true (horrible at text processing, dreadful with graphics, inconvenient process of compiling, linking etc.), but for number crunching Fortran remains just fine. Right from the earliest days, FORTRAN IV made it really simple to do complex arithmetic (with i imaginary), matrix algebra, and advanced statistical analysis. In this respect access to the freely available IBM SSP made numerical analysis a breeze. Less than 10 years later than PORTRAN saw the advent of widely available make-your-own microcomputers with the S100 bus allowing one to simply plug in whatever card was relevant to your needs. This was instant opportunity for someone bright to upstage FORTRAN forever. But it did not happen. Instead, FORTRAN-80 was born which ran on the Z80 chip. The entire software library of mainframes became available in your own private room. I don't find it in the least surprising that Fortran is still relevant today. It began life with a strong supporting library and has evolved and adapted to remain fit for purpose. Code portability is still an attractive feature. I am still programming 47 years later, and there is little need for me to write any new Fortran code because it is already all written and running perfectly. The only need is minor manipulation of new data for the same old programs. Finally, it be plainly absurd for me to think of using Fortran for other aspects of my academic work and research, such as interactive graphics, programming hardware, and writing simulation software. Fortran is not fit for purpose here, so I use Pascal and, reluctantly C.