XMIT wrote: ↑vometia wrote: ↑This programmer likes Model Ms. She also likes C and doesn't understand why some programmers like weird stuff like Fortran and Java and, God forbid, Pascal. And certainly not C++.
Heh. I've found that the more Java I have to write, the worse off I am. How do you feel about Python, then, or god forbid Perl?
I know I am late to the party and this is off topic but I couldn't help but chime in. My career primarily involves scientific computing so perhaps my perspective is a bit different from other software developers here, but I very much agree with this sentiment. The fact that it's still so prevalent even though there are many better alternatives now feels to me like a result of the lingering attachment to the JVM in enterprise applications. I especially dislike Java because I feel like it makes me waste too much of my development time writing needless boilerplate. I have seen it lead to wasteful habits on projects in other programming languages. For example in one of the python projects I am involved with, I noticed that there was a set of modules containing just one class, even though it's possible to define multiple classes in a single python file.
Since I do some big data analytics I also had a brief stint with Scala myself, and also didn't care for it too much. It fixes some of the verbosity issues with Java, but somehow it's even more needlessly complex. It very much strikes me as a language that was designed more for academics than actual production, but nevertheless it too has gained some traction thanks to its heavy use in big data frameworks on the JVM like Apache Spark. Above all else it feels somewhat clunky to use for any dataset that's more than two dimensional (something I must deal with all the time in my discipline area of earth science), which to me makes it far inferior to a language to Python which has very mature libraries like numpy and xarray for these use cases.
Fortran is still heavily used for scientific applications and high performance computing. In my field, this means most if not all weather forecasts are derived from the output of numerical models that were written in Fortran (90 or 77) as the prevailing attitude among scientists is that they generally feel content to use something that "just works", even if the code base is still ugly and (in some cases) decades old. In that sense, they aren't too different from us vintage keyboard enthusiasts who are fine with beige color schemes and large bezels rather than switching to a shiny, expensive newfangled custom! It's a stark contrast to some software developers on the other extreme who are willing to make their code backwards incompatible (ie, Python 2 vs 3) for the sake of "technical correctness"
.
