parser VS retro-httpaf-bench

I find straight forward, dedicated combinators much more readable and practical to use ie. for iterables (context where it makes a lot of sense) [0] example [1], runtime assertions (through refutations, which are much faster than combinators over assertions) [2], parser combinators for smallish grammars [3] etc.

In many cases vanilla/imperative js is more readable and terse, no need to bring functional fanaticism everywhere, just in places where it gives true benefits and in form that can be understood by peers.

Functional code can be beautiful and can also be unreadable/undebugable. Same with imperative code. It's great in js/ts you can pick approach where the problem is expressed more naturally and mix it at will.

[0] https://github.com/preludejs/generator

[1] https://observablehq.com/@mirek/project-euler

[2] https://github.com/preludejs/refute

[3] https://github.com/preludejs/parser

Parser combinators are great, we're using parser combinators in production, they are great ie. for typescript [0].

[0] https://github.com/preludejs/parser

I do. The benchmark results itself is here: https://aws1.discourse-cdn.com/standard11/uploads/ocaml/opti.... This comes from the OCaml multicore monthly news, the october 2021 edition: https://discuss.ocaml.org/t/multicore-ocaml-october-2021/882.... The benchmark's repo is here: https://github.com/ocaml-multicore/retro-httpaf-bench. However that image is not the whole story, and there's a bit more info here: https://watch.ocaml.org/videos/watch/74ece0a8-380f-4e2a-bef5.... In that video, the author says that the result vary depending on the load (sometimes Rust Hyper can end up above OCaml httpaf eio), that OCaml currently uses an io-uring backend while Rust doesn't, and that the results are for single core as previous OCaml implementations are single-core themselves.

I do feel that this benchmark is incomplete. I'd like it to see the results while using all of the cores of a machine, and I'd like to see different type of loads. I do think that the results are impressive: performance between Go and Rust is great. I do hope that it stays this way with multicore.

We don't compare against Go pervasively. Benchmarking across languages is hard generally, but here is a result on a specific benchmark comparing several versions of OCaml benchmarks against Go and Rust on a Http server benchmark: https://github.com/ocaml-multicore/retro-httpaf-bench/pull/1....

If there are suggestions to make the Go and Rust versions, please feel free to tell us how in the issue tracker.

multicore already faster than Go

Multicore is coming along, you can read the latest news here: https://discuss.ocaml.org/t/multicore-ocaml-june-2021/8134

In terms of performance, there is this paper https://kcsrk.info/papers/system_effects_feb_18.pdf where on a single core async OCaml and effect OCaml are close to Go's net/http, and there is also this project https://github.com/ocaml-multicore/retro-httpaf-bench but I haven't see any results from it.