LOloiseaujc·Sep 24, 2025
OP here. I actually love Julia and acknowledge its ecosystem can be fantastic for learning purposes. Just see the Computational Thinking course at MIT.
I'm not dismissing Julia. Actually, the first lines of my conclusions are
> In the end, when it comes to teaching the basics of numerical linear algebra, Python and Fortran are not that different. And in that regard, neither is Julia which I really like as well.
I feel like people got the impression that I'm saying they *should* use Fortran for teaching. That ain't the point, but maybe I did not convey it as clearly as I would have like. The point is : a programming language with strong typing, clear begin/end constructs, ensuring inputs to a function cannot be accidentally modified (otherwise it has to be a subroutine), etc actually makes it easier for the students to effectively *learn*
Read the rest
computational thinking rather than having to battle with syntax errors and strange intricacies of a general-purpose language. Fortran is just an example which turns out to be historically related to number crunching.
Unicode support and greek letters sure can be useful when presenting code snippets in your slides, but it essentially is syntactic sugar coating. And, unfortunately, many students have no idea how to spell greek letters (e.g. \to for \tau, \fi for \phi, etc) and just end-up loosing time on aesthetic details rather than focusing on the learning objective.
Finally, you may not know it but Fortran does have a package manager. Check out fpm (https://github.com/fortran-lang/fpm). It basically is just like Pkg for Julia, also using a toml manifest, can be installed via conda, makes sure things are reproducible across compilers and platforms, etc.