Bifoldable about cats HOT 8 CLOSED

Yep, I would like to see Bifoldable.

from cats.

It's worth mentioning that Foldable and Bifoldable are lawless. We admitted Foldable anyway due to its usefulness for adhoc polymorphism. Bifoldable occurs much less frequently in practice, so we might want to carefully consider its inclusion.

from cats.

Foldable kind of has laws: https://github.com/scalaz/scalaz/blob/series/7.2.x/scalacheck-binding/src/main/scala/scalaz/scalacheck/ScalazProperties.scala#L369

from cats.

Those only require that foldMap, foldLeft, and foldRight are mutually consistent. They only exist in Scalaz because Foldable can be implemented in terms of foldMap or foldRight. If you pick one as the abstract method and just derive the other, the law simplifies to a unit test of the derived method.

from cats.

Is there consensus here?

So far it seems like @ceedubs and @tpolecat would like to see this and @mpilquist is skeptical?

from cats.

I personally use Bifoldable from Scalaz on a regular basis and would like to have it.

Plus it gives rise to wildly useful things like separate on MonadPlus.

👍

from cats.

Re: Laws some quick Googling turned up

https://github.com/scalaprops/scalaprops/blob/master/scalazlaws/src/main/scala/scalaprops/scalazlaws/bifoldable.scala /cc @xuwei-k

and

https://github.com/ekmett/bifunctors/blob/master/tests/BifunctorSpec.hs#L169

from cats.

Re-reading this just now, I'm not skeptical of adding Bifoldable. 👍

from cats.