Hacker News Reader

Comment by ookdatnog | Hacker News Reader

> Haskell is a language for experimenting, success is not their goal and it would remove resources to actual research.

You rightfully point to Rust, which took a lot of inspiration from Haskell, but I think it's worth emphasizing just how much of the progress in programming languages in the last two decades was inspired by functional programming (many features were not invented in Haskell, but some like type inference were popularized by it).

For example: proper type inference, algebraic data types (enums in Rust) and consequently option types, pattern matching, property-based testing, immutability by default, parametric polymorphism (generics), ad-hoc polymorphism (type classes/traits/...), first class functions (very old idea but only recently common in mainstream languages), ...

View on HN

The Rust docs claim most things actually came from ML and OCaml: https://doc.rust-lang.org/reference/influences.html

They only mention Haskell's type classes and type families as inspiring a Rust feature directly (traits).

The original rust compiler was written in ocaml

It's kind of alluded to, but, while FP languages, as a broad category, were the progenitor of a lot of different PL ideas, Haskell ended up implementing many of them in a single place. It's fair to say "we were influenced by (this other language that originated an idea)", but it's also fair "and Haskell also has that feature". Do it enough times, and you start to see why the claim that Haskell is important; being a research language it's able to have all of these ideas implemented in it in pursuit of its design goals, where other FP languages pick and choose in pursuit of a different one.

I think it would be best to create a spin off of Haskell. A small subset with only the good parts of Haskell, few hand selected extensions, new prelude, with focus on performance, very easy tooling/build chain and use in production for businesses: "Production Haskell Lite".

I learned type inference in Caml Light, Haskell was still Miranda back then, and I bet any language designer of all major languages has similar backgrounds in regards to ML type inference.

Even type classe ideas can be found in CLU, ML, Objective-C, before the paper that gave origin to their adoption in Haskell.

For history buffs: https://en.wikipedia.org/wiki/Hindley%E2%80%93Milner_type_sy... was invented in 1969 (and rediscovered my Milner in 1978).

AFAIK, that was the basis of type inference in many of the non-mainstream languages before they branched off into their own more powerful type systems.

from lisp and ML, but yes, haskell is a white-coat language.

a white-collar language

no.

white-coat as in white lab coat, i.e. a research language.

Reply on news.ycombinator.com