gerbil VS fib

That's a strange one! Can you go to https://github.com/mighty-gerbils/gerbil/issues and post an issue outlining this with slightly more detail? What platform, C compiler, libc version etc.

New in std library: an S3 client, an SMTP client, SSL for Postgres (enables Heroku support), better CLI support (including multicall binaries), and plenty of module updates. Plus a few minor bug fixes.

See Gerbil Scheme homepage https://cons.io

Here is the discussion: https://github.com/mighty-gerbils/gerbil/discussions/1008

Gerbil Scheme < https://cons.io > just saw its release v0.18, with many usability and documentation upgrades, and a bunch of new functionality in the standard library. A "meta-dialect of Scheme with post-modern features", Gerbil layers a Racket-like module system (the best in the world by far) on top of Gambit Scheme (compiler that produces the fastest code), with lots of libraries as "batteries included" for production-level client/server code.

I'm more into Scheme than CL, but am aware of Coalton. My current lisp is Gerbil: https://cons.io which already has a type annotation system and will be enhancing it for the next major release (v19).

Another "post-modern" natively compiling Scheme is Gerbil Scheme [0]. It's seeing a lot of attention/enhancements lately, including some bounties to implement features.

[0]: https://cons.io

I've typically found rust to be about 100x faster than Python for the projects I've ported. This is consistent with https://github.com/drujensen/fib.

Here is a similar benchmark but includes compile times. https://github.com/drujensen/fib

How? Do you mean how do I know it's slow? Because it takes longer to run.

Write a typical computation such as Fibonacci in Java and Erlang/Elixir and compare. Fortunately someone has already done this.

Elixir is 3x slower than C and 2x slower than Java for this single thread example.

https://github.com/drujensen/fib

Apparently this upsets people for me to point this out. However, I did not say that Elixir was slow in general or a bad choice. It's an excellent choice for problems which suit parallelization or which require reliable, consistent performance.

Since the parent poster had commented that adding this multi-module dispatch would not be performant, I merely pointed out that the single thread peformance was already slow (as in, why worry too much about the performance cost of the multi dispatch suggestion).

Is there some advantage over repos that also show the results ?