You smile kindly. "Haskell is a dynamically-typed, interpreted language."
4
2
32
Conversation
Funnily enough the meta-language of the Haskell type system is more similar to a simply typed lambda calculus, there are... *clears throat* "kinds" of types. wiki.haskell.org/Kind
1
1
Of course. "Maybe" is clearly of a different character than "Int", but both are types.
1
1
Indeed, they are both types, just like 42 and True are both values. But they are different "kinds" of type * -> * and * resp. You can get a static "compile time" kind error, which is like a type error "one level up" so to speak. i.e. the type annotations themselves are incoherent
1
One of the things my students of Haskell consistently have trouble with is understanding why a declaration like "instance Monoid Either where..." isn't allowed.
1
Maybe the mental model error is related to the inelegance that the kind system is not polymorphic. e.g. you can 'map' a unary or binary function, so why not Monoid Either?
1
1
Would be neat to be able to curry kinds. 🤔
1
1
Yeah the special rules around type level programming in haskell is annoying. I think the dependent haskell stuff aims to make it a bit less janky? I dunno though.
1
1
I do like that they are moving to `Type` instead of `*` for the type of types - would have made this stuff easier to learn at first! My brain always got really confused seeing stuff like `* -> *`.
3
1
But yeah, in spite of this, I think the quoted article is talking about how if you use `Type` for all your type-level programming it ends up being like you're using `Any` everywhere.