4 ms·

Fortran is suddenly hot again. But its future is still far from certain

24 points by hkmaxpro 5y ago

jschveibinz 5y ago
FORTRAN quotes: “You can tell how far we have to go, when FORTRAN is the language of supercomputers." — Steven Feiner. "FORTRAN, the infantile disorder, by now nearly 20 years old, is hopelessly inadequate for whatever computer application you have in mind today: it is now too clumsy, too risky, and too expensive to use." — Dijkstra, circa 1970. "FORTRAN was the language of choice for the same reason that three-legged races are popular." — Ken Thompson, "Reflections on Trusting Trust" "In the good old days physicists repeated each other's experiments, just to be sure. Today they stick to FORTRAN, so that they can share each other's programs, bugs included." — Dijkstra (1930—2002). "Consistently separating words by spaces became a general custom about the tenth century A.D., and lasted until about 1957, when FORTRAN abandoned the practice." — Sun FORTRAN Reference Manual "FORTRAN is not a flower but a weed — it is hardy, occasionally blooms, and grows in every computer." — Alan J. Perlis. "You can create bad Fortran in any language." —? “I don't know what the programming language of the year 2000 will look like, but I know it will be called FORTRAN.” — Dijkstra “Computers are useless. They can only give you answers.” - Picasso From: http://www.csl.mtu.edu/cs5311.ck/www/quote.html http://www.csl.mtu.edu/cs5311.ck/www/quote.html and https://www.gdargaud.net/Humor/QuotesProgramming.html https://www.gdargaud.net/Humor/QuotesProgramming.html
- m_mueller 5y ago
  This is kinda like judging C by how bad Algol was. Fortran 90 and its successors should be looked as a separate language from the old fixed format Fortrans, and they are the currently IMO the best we got for supercomputing purposes (i.e. I'd prefer it over C/C++, in that specific domain). That being said, I hope Julia pushes through as a replacement. Note that I'd never advocate for Fortran in any capacity outside of that domain, e.g. for system programming.
  - TheRealKing 5y ago
    The problem with Julia is that there is so much hype around it, likely because of the for-profit company behind Julia. I see a non-negligible number of false claims and false promises on Julia's performance on the web every now and then. and when people test it for real production code they get disappointed with it.
    - cbkeller 5y ago
      I don't think that's particularly what's going on. I think it's more just that Julia has a relatively small but incredibly zealous group of users. If Julia solves your personal version of the two language problem, and especially if dispatch-oriented programming happens to be a good fit for the problems you solve and the way you like to program, you're probably going to think it's the best thing since sliced bread. If not, on the other hand, you may not think it's anything particularly special.
    - amkkma 5y ago
      "I see a non-negligible number of false claims and false promises on Julia's performance on the web every now and then. and when people test it for real production code they get disappointed with it." Can you cite one or more specific examples of this? I'd like to see what you mean. Edit: You seem to be a Fortran fan
anotha1 5y ago
You can't ignore context. Fortran is a very strong language for a specific application.
corty 5y ago
FORTRAN is still the performance normal any numerics and linear algebra language is measured against. Up to now, most fail or require very ugly code to get up to FORTRAN levels of performance. Others cheat and just link FORTRAN libraries to do the heavy lifting.
hilbert42 5y ago
When I was learning Fortran I thought there were some ugly features about it especially its I/O - format statements etc., but what I thought was irrelevant (even to my lecturers). What people think of it doesn't matter much (most of its critics aren't regular users anyway). It seems to me the only real issues with Fortran are whether the huge legacy of the nearly 70 years of engineering and scientific programs and well-tested subroutines will continue to be used or are declared out of date/useless — and whether the many engineers and scientists who now use supercomputers stop using it and change to something else. I'd put a small wager on that it will likely be around for a while yet (it might even be the first working language to make it to 100).
ale42 5y ago
"as scientists are flocking to newer languages such as Python or Julia."... well, guess what? SciPy is partly written in... Fortran (and C/C++)! By its own nature, Python can't be as fast as Fortran. That's why most Python libraries doing heavy computation are implemented in Fortran and/or C/C++...

24 points by hkmaxpro 5y ago

FORTRAN quotes: “You can tell how far we have to go, when FORTRAN is the language of supercomputers." — Steven Feiner. "FORTRAN, the infantile disorder, by now nearly 20 years old, is hopelessly inadequate for whatever computer application you have in mind today: it is now too clumsy, too risky, and too expensive to use." — Dijkstra, circa 1970. "FORTRAN was the language of choice for the same reason that three-legged races are popular." — Ken Thompson, "Reflections on Trusting Trust" "In the good old days physicists repeated each other's experiments, just to be sure. Today they stick to FORTRAN, so that they can share each other's programs, bugs included." — Dijkstra (1930—2002). "Consistently separating words by spaces became a general custom about the tenth century A.D., and lasted until about 1957, when FORTRAN abandoned the practice." — Sun FORTRAN Reference Manual "FORTRAN is not a flower but a weed — it is hardy, occasionally blooms, and grows in every computer." — Alan J. Perlis. "You can create bad Fortran in any language." —? “I don't know what the programming language of the year 2000 will look like, but I know it will be called FORTRAN.” — Dijkstra “Computers are useless. They can only give you answers.” - Picasso From: http://www.csl.mtu.edu/cs5311.ck/www/quote.html http://www.csl.mtu.edu/cs5311.ck/www/quote.html and https://www.gdargaud.net/Humor/QuotesProgramming.html https://www.gdargaud.net/Humor/QuotesProgramming.html

m_mueller 5y ago
This is kinda like judging C by how bad Algol was. Fortran 90 and its successors should be looked as a separate language from the old fixed format Fortrans, and they are the currently IMO the best we got for supercomputing purposes (i.e. I'd prefer it over C/C++, in that specific domain). That being said, I hope Julia pushes through as a replacement. Note that I'd never advocate for Fortran in any capacity outside of that domain, e.g. for system programming.
- TheRealKing 5y ago
  The problem with Julia is that there is so much hype around it, likely because of the for-profit company behind Julia. I see a non-negligible number of false claims and false promises on Julia's performance on the web every now and then. and when people test it for real production code they get disappointed with it.
  - cbkeller 5y ago
    I don't think that's particularly what's going on. I think it's more just that Julia has a relatively small but incredibly zealous group of users. If Julia solves your personal version of the two language problem, and especially if dispatch-oriented programming happens to be a good fit for the problems you solve and the way you like to program, you're probably going to think it's the best thing since sliced bread. If not, on the other hand, you may not think it's anything particularly special.
  - amkkma 5y ago
    "I see a non-negligible number of false claims and false promises on Julia's performance on the web every now and then. and when people test it for real production code they get disappointed with it." Can you cite one or more specific examples of this? I'd like to see what you mean. Edit: You seem to be a Fortran fan

anotha1 5y ago

You can't ignore context. Fortran is a very strong language for a specific application.

corty 5y ago

FORTRAN is still the performance normal any numerics and linear algebra language is measured against. Up to now, most fail or require very ugly code to get up to FORTRAN levels of performance. Others cheat and just link FORTRAN libraries to do the heavy lifting.

hilbert42 5y ago

When I was learning Fortran I thought there were some ugly features about it especially its I/O - format statements etc., but what I thought was irrelevant (even to my lecturers). What people think of it doesn't matter much (most of its critics aren't regular users anyway). It seems to me the only real issues with Fortran are whether the huge legacy of the nearly 70 years of engineering and scientific programs and well-tested subroutines will continue to be used or are declared out of date/useless — and whether the many engineers and scientists who now use supercomputers stop using it and change to something else. I'd put a small wager on that it will likely be around for a while yet (it might even be the first working language to make it to 100).

ale42 5y ago

"as scientists are flocking to newer languages such as Python or Julia."... well, guess what? SciPy is partly written in... Fortran (and C/C++)! By its own nature, Python can't be as fast as Fortran. That's why most Python libraries doing heavy computation are implemented in Fortran and/or C/C++...