FromFoldable and FromFoldable1

[fommil]

Maybe I'm not fully grokking Unfoldable (which should probably have a ticket) but I found myself needing these trivial little things

trait FromFoldable[F[_]] {
  def fromFoldable[G[_]: Foldable, A](g: G[A]): F[A]
}
trait FromFoldable1[F[_]] {
  def fromFoldable1[G[_]: Foldable1, A](g: G[A]): F[A]
}

with mostly trivial implementations that all our data types could have and could have a CanBuildFrom implicit for legacy reasons. This allows the removal of a lot of methods across the board that are effectively Foldable.toAHardCodedDataType

I guess what we lose is the ability for Foldable[IList] to have an optimal toIList method, and such, which is maybe an argument to retain some hardcoded methods on Foldable.

Possibly related to #1452 and #1448

[jbgi]

@fommil did you have a look at Reducer (and maybe Generator)? I think they could help.

In particular I think we could add the following to Foldable:

  def reduceTo[A, M](fa: F[A])(implicit M: Reducer[A, M]): M = ???
  def to[A, F[_]](fa: F[A])(implicit M: Reducer[A, F[A]]): F[A] = reduceTo[F[A]]

But maybe there is already a better way that I did not see.

[Jacoby6000]

@jbgi interestingly, it looks like ReducerOps has almost those exact methods hidden inside a sealed abstract class, UnfoldTo.

scalaz/core/src/main/scala/scalaz/syntax/ReducerOps.scala

Line 47 in 1a01e00

sealed abstract class UnfoldTo[C] {

[fommil]

I love deleting hacky code that I wrote 😄 I'll take a look and report back tonight.

We should definitely get these added to Foldable. For 7.3 I'm all in favour of removing the stdlib collection methods on Foldable and requiring people to "reduce" to a LiterallyCantEvenBuildFrom

[fommil]

I think I need a Reducer1 (for non empty things)

[jbgi]

@fommil yes, could be worth solving #1038 for 7.3

[fommil]

https://hackage.haskell.org/package/reducers-3.12.2/docs/Data-Semigroup-Reducer.html

[fommil]

@jbgi I'm still not getting Reducer... can you please show me an example of using it to create a NonEmptyList from a Foldable?

[jbgi]

Yeah, will try to put together a Reducer1 over the holidays

[fommil]

@jbgi much appreciated, this is needed by scalaz-deriving (for the N-arity case) and my hacky FromFoldable1 is... hacky.

[fommil]

I'm going to close this in the hope that reducers will solve the issue of creating a data structure from a list / nel