[Insanity]

I'm thoroughly surprised to see that Fortran is still being updated (with an upgrade still in the works for next year apperantly). I had no idea.

I'll give it a shot as it seems interesting to at least have worked in it a bit, but I do wonder, where is Fortran used these days?

A quick glance at Tiobe[1] shows that it scored a bit higher than Haskell, Scala and Kotlin. Can anyone explain me why?

[1]: https://www.tiobe.com/tiobe-index/

[rootbear]

I work at NASA Goddard Space Flight Center and I can tell you that Fortran is indispensable. It is the primary language used in numerical simulations, such as climate, stellar dynamics, cosmology, etc. This is partly due to inertia; scientists don't like to spend time learning new programming languages. Also, the various Fortran libraries used in this work are thoroughly understood and no one wants to change them or port them. Most of the code here is in Fortran 95, which may look odd, but has all the things a reasonable "Structured Programming" language should have. Later versions added some Object Oriented features but I don't think they've caught on much around here.

I don't personally work in Fortran, but I'm not repelled by the idea. The thing that would attract me to it, as a former computer graphics programmer, is the decent support for arrays. C99 is better than C89, but it's not where Fortran is. Sure, C++ can do it with classes but, goodness me, that is one complex language. One pays a hefty price if all one wants is decent array handling. And speaking of complex, Fortran has had complex numbers for decades. I've never used complex numbers in C99, so I don't know how they compare. I'm hearing murmurs of moving some things to Rust, when it matures, but I think the jury is still out on that.

I always like to include the quote below in discussions like this:

“I don’t know what the programming language of the year 2000 will look like, but I know it will be called FORTRAN.” – Charles Anthony Richard Hoare, circa 1982

[azag0]

Personal experience: My general language-to-go is Python, and I'm very fond of Rust. But Fortran has no competition in numerical computing at native speeds in terms of convenience. I was just recently rewriting a small Fortran library (that I usually but not always call from Python) to C, and it was a pain. So, as a result of that, Fortran is still used heavily in scientific computing. (On the other hand, Fortran is a pain for anything else than numbers and arrays.)

It is also the only language I'm aware of that has a special syntax for distributed memory [1], making parallel distributed computing potentially extremely convenient.

[1] https://gcc.gnu.org/wiki/Coarray

[nickpsecurity]

What do you think about Julia language? It's a modern take on a language for number crunching that has support for C and Python code. Also uses some Fortran libraries that are optimized.

https://julialang.org

Far as distributed, see Cray's Chapel and Taft's Parasail for languages trying to improve on that. Chapel's competition was IBM X10 and Fortress languages. Im not sure if those two are maintained any more, though.

[kdelok]

In general the performance of C vs Fortran is a moot point. However, for numerical performance, Fortran has (at least) two major advantages that I can see, both of which are about usability:

- In-built support for multi-dimensional arrays and operations on arrays. This makes it so much easier to work with numerical datasets than in C.

- BLAS and LAPACK are the two foremost packages used for linear algebra. They're written in Fortran, so it's just a bit easier to use them, rather than their C APIs. I've had unpleasant experiences debugging things across the boundary between C and Fortran.

[simplicio]

I did my dissertation in astrophysics a few years ago using simulations I programmed in Fortran. As others have said, its efficient for number crunching, there's a lot of legacy code in scientific fields and most older Profs are familiar with it.

Also, as a language, there isn't much to it. If you're already familiar with the basics of programming, you can learn most of what there is to know about Fortran 77 in an afternoon, and pick up later additions to the language as you go. In physics, where the focus is getting grad-students to a point where they can do research quickly, having a language they don't have to spend a month figuring out is an advantage.

[jcvernaleo]

My dissertation in astrophysics was pretty heavy in FORTRAN too (hydrodynamics simulations). Legacy code, simple language, and really fast compilers.

[nickpsecurity]

Lots of optimized libraries used in scientific computing are done in Fortran with the language itself making some optimizations easier just because it's more conservative. All the HPC vendors selling supercomputers and such have Fortran on them. The language itself has been updated periodically, too. Add a legacy effect of people just using it for a long time and you get a language that isn't going away.

[kbenson]

> I do wonder, where is Fortran used these days?

Not that I actually know, but what I hear is that it's still used for number crunching, and it's actually faster than C in some respects unless you make heavy use of C99 features. That is, very generally it doesn't support a few features of C that makes it easier to optimize some constructs, nut that C now has keywords which can signal the compiler appropriately so the same optimizations can now happen in C.[1]

I just looked that up from memory of prior discussions here though, so if someone with real experience chimes in, take them over me.

1: https://stackoverflow.com/questions/146159/is-fortran-easier...

[love2lol]

It's a language whose syntax makes it easy to pass around multi-dimensional arrays, bounds checking is a standard feature on most compilers, doesn't have the same aliasing problems as C so compilers can do more aggressive optimizations and it compiles directly to your native hardware's instruction set (no vms/sandboxes). Why wouldn't you want that for cpu bound numerical code?

Also, the PG (now nVidia) Fortran compiler makes it very easy to make your code run on a gpu with only adding some attributes to variables and some pramgas.

[rpeden]

And COBOL scored higher than Fortran! Interestingly enough, it's still being updated too. COBOL 2014 is the latest ISO standard.

[pmbailey]

FORTRAN is deeply ingrained in engineering culture. A lot of legacy (and some new development) is written in it. Though, this is slowly shifting with the prevalence of MATLAB in universities.

[Mikeb85]

Because it's an easy language to pick up and use, and so well optimized that it'll always be blazing fast and parallelized.

[yarrel]

A lot of very solid code has been written in it over the years and continues to be used as libraries accessed from e.g. R.

[dagw]

where is Fortran used these days?

I see a lot or Fortran still being written in various parts of climate and earth science.