Fix #617: Make Resource covariant in F and A

[neko-kai]

Following the discussion in #617 this simply makes Resource covariant. Binary compatibility is not affected as non-generic parts of the method signatures are unchanged.

[neko-kai]

Strange Travis errors, not related to sbt:

bundler's executable "bundle" conflicts with /home/travis/.rvm/rubies/ruby-2.6.0/bin/bundle

Overwrite the executable? [yN]  

No output has been received in the last 10m0s, this potentially indicates a stalled build or something wrong with the build itself.

Check the details on how to adjust your build configuration on: https://docs.travis-ci.com/user/common-build-problems/#build-times-out-because-no-output-was-received

The build has been terminated

Maybe the build needs restarting?

[rossabaker]

That's yet another gift from our Ruby dependency. It's fixed in #729.

[neko-kai]

@rossabaker Thanks for pointing that out!

[neko-kai]

MiMa failures seem to be false positives, lightbend-labs/mima#363

[rossabaker]

Seems reasonable to me, but I'd like to hear from @mpilquist or @SystemFw, who as noted in the linked ticket did similar in fs2.

[rossabaker]

Note the Ruby change will conflict with #733, which I think is a better solution, but it will be an easy merge for whoever goes second.

[LukaJCB]

Do we know how this will affect type inference? I can see some extra type annotations in the changes that weren't needed before.

[neko-kai]

@LukaJCB
Seems like F parameter infers worse in Resource.pure when import cats.implicits._ is present, because many high-priority candidates for Applicative cause 'ambiguous implicits' error:

Error:(468, 37) ambiguous implicit values:
 both method catsStdInstancesForEither in trait EitherInstances of type [A]=> cats.MonadError[[β$0$]scala.util.Either[A,β$0$],A] with cats.Traverse[[β$1$]scala.util.Either[A,β$1$]]
 and value catsStdBimonadForFunction0 in trait Function0Instances of type => cats.Bimonad[Function0]
 match expected type cats.Applicative[F]
      case Right(a) => Resource.pure(a)

fs2 avoids that in Stream.apply by returning in essence Stream[Nothing, A], but Resource's pure is not free, so we can't do the same.

Personally, I use neither cats.implicits nor Resource.pure, and I expect that aside of .pure the inference should be on par with fs2.

[djspiewak]

This is awesome, thank you so much! I want to sort out the false positives on mima so we don't break the master build. I'm not super concerned about the source compatibility issue with pure; it's unfortunate, but not the end of the world, and maybe fixable later before release.

[joroKr21]

Hmm that specific problem (type inference of pure) seems to be solved by scala/scala#8582, although for reasons other than implicit scope. Tried quickly in the REPL:

scala> class Applicative[F[_]]
class Applicative

scala> implicit val apList: Applicative[List] = new Applicative[List]
val apList: Applicative[List] = Applicative@2ad7bd26

scala> implicit val apVec: Applicative[Vector] = new Applicative[Vector]
val apVec: Applicative[Vector] = Applicative@149d7cc6

scala> class Resource[+F[_], +A]
class Resource

scala> def resource[F[_], A](implicit ap: Applicative[F]): Resource[F, A] = new Resource
def resource[F[_], A](implicit ap: Applicative[F]): Resource[F,A]

scala> val x: Resource[List, Int] = resource
val x: Resource[List,Int] = Resource@33f81280

I wonder what that means about enhancing lower bounds (but probably not safe in Scala 2 due to LUBs).

[neko-kai]

@joroKr21
That's great, another quality of life boost for variance connoiseurs brought on by your patch, thanks! 🙏

[neko-kai]

@djspiewak I've added exclusions for MiMa, the build passes now.

[djspiewak]

Sorry for the long delay on this. Thanks for all your hard work, @neko-kai!