Propose an implicit val for the default instance for Console[IO] to the user

[guizmaii]

Hi @gvolpe,

While working on the update of advanced-http4s, I remarked that you don't propose an implicit val for the default instance of Console[IO].

It's useful when you work with code like this:

  def stream[F[_]: Console]: F[Unit] = ...

  override def run(args: List[String]): IO[ExitCode] = {
    import cats.effect.Console.implicits._

    stream[IO].as(ExitCode.Success)
  }

WDYT ?

[gvolpe]

I can't remember exactly why we didn't make that val io: Console[IO] = SyncConsole.stdio[IO] implicit but I'm thinking it should be. So there would be no need to introduce some extra object. @kubukoz thoughts?

[kubukoz]

I thought Console isn't a typeclass and as it's and algebra it'd be better not to have an implicit instance (one can create an IO Console for tests and accidentally not use it). I like the proposed solution with the implicits object. That'll require an import to bring in the instance, and the above issue will be less common.

[gvolpe]

I think that was the reason 👍 . I like the idea of implicits.io._

[guizmaii]

@gvolpe the import will have this form: import cats.effect.Console.implicits._ or this one import cats.effect.Console.implicits.ioConsole but not implicits.io._

[gvolpe]

@guizmaii why not import cats.effect.Console.implicits.io? Having "console" twice seems kind of redundant.

[kubukoz]

That's just for the global namespace thing in think. Or does it not apply to implicit parameters (maybe just conversions)?

[guizmaii]

@gvolpe
As @kubukoz said:

ioConsole, which would be even better because implicits have kind of a global namespace and io will potentially clash very easily

It's safer.

[gvolpe]

Right on, sounds good to me! Will merge if no objections.

[guizmaii]

Thanks for your reviews @gvolpe @kubukoz :)

[gvolpe]

Thank you for putting up the work @guizmaii , I just updated the guide -> https://gvolpe.github.io/console4cats/guide.html