Learning FORTRAN In the Modern Era

Question

I\'ve recently come to maintain a large amount of scientific calculation-intensive FORTRAN code. I\'m having difficulties getting a handle on all of the, say, nuances, of a

Answer 1

I started on Fortran IV (WATFIV) on punch cards, and my early working years were VS FORTRAN v1 (IBM, Fortran 77 level). Lots of good advice in this thread.

I would add that you have to distinguish between things done to get the beast to run at all, versus things that "optimize" the code, versus things that are more readable and maintainable. I can remember dealing with VAX overlays in trying to get DOE simulation code to run on IBM with virtual memory (they had to be removed and the whole thing turned into one address space).

I would certainly start by carefully restructuring FORTRAN IV control structures to at least FORTRAN 77 level, with proper indentation and commenting. Try to get rid of primitive control structures like ASSIGN and COMPUTED GOTO and arithmetic IF, and of course, as many GOTOs as you can (using IF-THEN-ELSE-ENDIF). Definitely use IMPLICIT NONE in every routine, to force you to properly declare all variables (you wouldn't believe how many bugs I caught in other people's code -- typos in variable names). Watch out for "premature optimizations" that you're better off letting the compiler handle by itself.

If this code is to continue to live and be maintainable, you owe it to yourself and your successors to make it readable and understandable. Just be certain of what you are doing as you change the code! FORTRAN has lots of peculiar constructs that can easily trip up someone coming from the C side of the programming world. Remember than FORTRAN dates back to the mid-late '50s, when there was no such thing as a science of language and compiler design, just ad hoc hacking together of something (sorry, Dr. B!).