Add foldMapK and reduceMapK

[LukaJCB]

No description provided.

[kubukoz]

foldMapK's signature would look something like this, right?

def foldMapK[G[_], A, B](fga: F[G[A]])(f: A => B)(implicit G: MonoidK[G]): G[B]

(and similarly for reduceMapK, with the change MonoidK -> SemigroupK)

[kubukoz]

hmm, I think I wouldn't be able to implement that without an Applicative[G], which is probably not the point, so I must've got the signature wrong

[kubukoz]

def foldMapK[G[_], A, B](fga: F[A])(f: A => G[B])(implicit G: MonoidK[G]): G[B] =
  foldMap(fa)(f)(G.algebra)

sounds more like it, WDYT?

[LukaJCB]

Yes A => G[B] is what I had in mind :)

[kubukoz]

Cool, I'll work on it then

[kubukoz]

I'm wondering how we can test this.

Should there be a new law that says foldMapK(fa)(f) <-> fa.map(f).foldK? That'd require a Functor[F] in the first place, and there are probably types out there that have a Foldable but not a Functor.

Alternatively, we could have tests for some concrete instance that the equivalence holds, or have a law that reuses what's in the implementation (although I doubt that's the right way to go).

WDYT?

[LukaJCB]

I think a better law would be foldMapK(fa)(f) <-> foldMap(fa)(f)(MonoidK[G].algebra) :)

[kubukoz]

Could be 😅 that's actually what I meant by having a law that reused the implementation we're going to have - but I guess it would suffice for testing people's implementations if they decide to override

[kubukoz]

I'm working on the laws, apparently adding a new type parameter (G[_]) breaks tens of other tests, so bear with me

[LukaJCB]

You could use Id like is done for foldM :)

[kubukoz]

You're telling me now 😂

I already did it for G[_] and passed Option everywhere (as is the custom that I saw in the use sites), in case it's too much to add I can hardcode Id. I'll open a PR now so you can see what it looks like