The oil and gas industry still runs on Fortran code written many decades ago. The mathematicians and physicists got things right way back then in Fortran, no need to rewrite it.
I feel this, my field (orbital dynamics) has a lot of heritage fortran that at this point I think will never go away. Much of it has a proven history, but there are large swaths of it that the original authors have long since died and no one knows the innards anymore.
I know, if it aint broke... but it feels like something has been lost if we are using code that no one alive knows what it is doing. I suppose that is the AI future.
Every time I've heard about a present-day Fortran code base it's because 1) Fortran is really good at those sorts of problems, and 2) lock-in because it's too hard to rewrite to another language, with little career benefit when you could be adding new features.
Because modern Fortran has all the goodies as other languages, first class support for CUDA, and a performance Python will never achieve, all coupled with 70 years of production code.
Also lets not forget that the beloved C is only about 10 years younger.
I know, if it aint broke... but it feels like something has been lost if we are using code that no one alive knows what it is doing. I suppose that is the AI future.
Also lets not forget that the beloved C is only about 10 years younger.