@AmenZwa I was thinking about your https://amenzwa.github.io/stem/PL/FortranModernisation/ .

[screwlisp@gamerplus.org]

@AmenZwa I was thinking about your https://amenzwa.github.io/stem/PL/FortranModernisation/ . In my small experience, research labs at at least some universities have been deeply opposed to fortran, and supportive of writing what should have been and almost was fortran in python or non-fortran-using matlab. Switching to Fortran being above their paygrade, and inimical to their business community besides.

Though, I do agree with you that if you want to write some algebra, writing fortran directly is suitable language for that.

[amenzwa@mathstodon.xyz]

@screwlisp You stated in a few words what I tried to say in that lengthy article. Right Fortran, even the 2023 version, suffers from image problem amongst executives, technologists, and students alike. But there is a very small group of ardent Fortran devotees in a number of federal government agencies (some of them are my clients), several engineering professors (some of them are my classmates), and a few engineering students (my friends’ mentees). All of them—their numbers are few, but their devotion ample—rely on supercomputers, and they all swear by PGAS to implement their simulations.

One of my gov clients is indeed in trouble. They’re still running F77 applications on HP-9000s. The hardware parts are drying up, the software has rotted, and the wetware is retiring or dying.

[vnikolov@ieji.de]

Yes...

And it has actually been pointed out a number of times
that Fortran is _still_ superior to many, if not most,
other programming languages for numeric computations.
In a number of ways.

@AmenZwa @screwlisp

[amenzwa@mathstodon.xyz]

@vnikolov
Right on! Yes, FORTRAN is respected in narrow, but deep, subarea of our field, namely scientific and engineering parallel numerical computations. Likewise, COBOL is still used in major financial institutions and federal agencies that regulate those institutions. For some thirty years or so, this lot had tried, repeatedly, to dislodge COBOL by switching to C++, Java, Python, and the like, but failed, repeatedly. It turned out that nothing beats COBOL's transactional throughput, when it is run on a modern mainframe mated to a modern I/O fabric. But like the FORTRAN lot, COBOL shops, too, are on shaky grounds. Both camps are facing an alarming shortage of wetware, due to retirements, and due to the drying up of the replacement pipeline.

On the contrary, in the present company, it'd be superfluous to harp about how LISP is thriving and steadily attracting new, curious programmers, after almost seven decades. It could, at least in part, be due to the recent (since around 2010s) resurgence of FP in IT.

Most EE kids today came into the uni armed with Python adoration, and most CS kids came into the uni believing that they're JavaScript demigods. And unlike in our days, when most kids try to learn at least a few different languages, kids these days are perfectly happy with their one true language, be it Python or JavaScript, because the language coupled with its ample, modern ecosystem, are essentially "complete" kit. This single-language scenario is productive and less frustrating, for sure. But it gives these kids a sense of unwarranted superiority, which will bite them on their posteriors, when that "currently number 1" language wanes in popularity.

@screwlisp

[amoroso@oldbytes.space]

@AmenZwa I thought @arclight might be interested in this conversation.

@vnikolov @screwlisp