Jon Pretty

It's the abstracting over that's the problem for Scala, rather than the use (which is great when the type is concretely known)... the main problem is that for-comprehensions allow more general types than can be easily represented by an OOP-style typeclass interface...

Adding a F[_]: Monad type constraint and a cats.implicits._ import, or defining the instance by implementing a trait is easy enough. The problem is there’s no standard way to do this, so libraries roll their (incompatible) own, or just avoid altogether

I think part of the problem is that a dependency on the whole of Cats or Scalaz, if all you want is a way to abstract over monad-like things, is too much... which is why people roll their own...

This might be only a psychological barrier, I suspect cats might be one od the smaller dependencies (measuring in bytes). And no transitive ones, too

Yes, definitely psychological to some degree. But also, if a dependency is very common dependency and has many versions which other libraries may use, that's a large surface area of potential conflict. I think that's where the real cost lies, rather than the size in bytes.

If it's that common, isn't (at least the "core" part of it) a good candidate for std lib then? That would significantly reduce the potential area of conflict :)

It could... but then it moves to an 18-24 month release cycle. ;)

That's the point, you don't want it constantly breaking under you.