what's the point of using Rust, Go, etc. for speed when

you can just write Python all the way down and write C++ modules (Boost::Python) for the slow parts as needed in the same way numpy works.

if you need everything to be fast and scalable, then just use Java.

I don't see any use case for Go, Rust, Haskell, and other compiled shitlangs that aren't C/C++.

Python isn't fast
Java is bloated and not fast

Go and Rust have a big advantage over C, C++ and Python: they allow easy cross-compilation out-of-the-box. building something that is just Linux and (native, not cygwin) Windows in C or C++ requires either two sets of makefiles (and code that doesn't rely on non-standard additions to your're are compiler of choice) or shit like CMake/autoconf, and you still need to compile it on a target machine. Python programs need to be either compiled with Cython or some library like PyInstaller but IIRC you still can't cross-compile because some Windows-specific files are required. also, I think that dependency management in those libraries was a bit of a pain, but I might be mistaken.

Haskal is sufficiently different from all the other languages that the biggest advantage is just using Haskal. I'm not a hasakler so I can't comment on that though, when it comes to FP I'd rather use a lithp - or, if typefaggotry is desired, OCaml

>>2

Java is bloated and not fast

that's not really true anymore, JVM has some really great JIT and Android's AOT is also quite nice - so if you write it sensibly (which more or less means 'not the typical over-abstracted labyrinth-oriented Java way'), it's gonna be decently fast. cold startup time, RAM usage and non-deterministic GC pauses are still going to be a thing, but it's manageable, especially if you know about the internals (which theoretically you shouldn't have to, but it's unavoidable - writing highly performant C, C++ or Fortran still requires you to understand both the compiler heuristics used for optimizations and some hardware-specific shit like caches).

I've seen a decently performant forex system built on Java, but the code definitely didn't look like your're are average enterprise system. personally, I wouldn't use this language for HPC because of the GC, but you'd still be surprised how good it can be in the hands of an experienced programmer.

C++: spending 10 hours to make your program take 3 seconds less to start up

Python: spending very little time to develop a program that is ``slow'' in terms of CPU cycles

What's more important: CPU time or developer time? I'd say the latter.

>>5
depends on the use case. you wouldn't write device drivers in Python

>>6
writing device drivers is for boomers

>>7
your're are an anus

>>8
better to be an anus than a boomer

>>9
that's not mutually exclusive!

bad: being an anus
worse: being a boomer
worst: beeing a boomer anus
best: checking my dubs

I don't see any use case for Go, Rust, Haskell, and other compiled shitlangs that aren't C/C++.

There is no compiled language but C and C++ is its preprocessor!

But numpy is slow as fuck.