3 ms·
Curious - there is one unnoticed comment on the thread that reminds me of my experience: > I feel the opposite. Was trying some number crunching, ok net doesnt
by vitorsr 2y ago
Curious - there is one unnoticed comment on the thread that reminds me of my experience:
> I feel the opposite. Was trying some number crunching, ok net doesnt have an fft.
> What about digital filters... well... lacking
> Displaying scientic graphs, like matplotlib. Nope, there might be a commercial package, but I didnt get a budget.
As someone trying to develop numerical “enterprise” applications in the most boring, maintainable and longevous stack as possible, am I to write and maintain my own foreign function interfaces?
- neonsunset 2y agoI don’t think this is a practical reality though. There is very good math and SIMD APIs for solving these kinds of problems natively, and finding decent bindings is just a matter of using search on nuget and GitHub. And given how dynamic the progress has been, “not possible” changing to “first class” can happen within a year or two.
- aragilar 2y agoThe question is then, why not use the native language, or a language where the binding are well-established (and not buggy-as-hell, as is the case with JS)?