Port to Cats.

[sellout]

This is done in a style that attempts to minimize the diff from the Scalaz version, for as long as a Scalaz branch is supported.

[julien-truffaut]

Thank you so much Greg, I will try to review it and check why tests are failing this weekend.

[sellout]

The test that’s failing here is Map.Each.get what you set, which seems to be simply reversing the List that is the set value somehow. And it’s only failing on the JVM (although locally it seems to be failing on only JS ¯\_(ツ)_/¯).

Also, Travis still seems to be running 2.10 and Scala Native jobs, even though I thought I disabled those in both .travis.yml and build.sbt.

[julien-truffaut]

I added the following unit test:

  test("each getAll") {
    val x = Map(1 -> "abc", 2 -> "cde")
    val e = each[Map[Int, String], String]
    e.modify(_.reverse)(x) should be (Map(1 -> "cba", 2 -> "edc"))
    e.getAll(x) should be (List("abc", "cde"))
    e.getAll(e.modify(_.reverse)(x)) should be (List("cba", "edc"))
  }

and I get:

[info] - each getAll *** FAILED ***
[info]   List("edc", "cba") was not equal to List("cba", "edc") (MapSpec.scala:18)

I suppose it comes from the Map Traverse instance in cats, it does not require the key to be ordered like scalaz does.

[julien-truffaut]

I reported the issue in cats: typelevel/cats#1831

[tpolecat]

So, I think the problem is that your test is assuming something that isn't necessarily true. The equivalence relation Eq isn't strong enough to prove congruence; a === b doesn't imply that f(a) === f(b) which is kind of at the root of this problem; sets a and b may be === but a.toList =!= b.toList, which is irritating but legal. So maybe all that needs to happen is to weaken the test expectations?

[julien-truffaut]

I changed the target branch to cats as I prefer to make the first few releases with cats dependency using a different artifact (e.g. monocle-core-cats).

Thank you very much Greg, it is really an amazing work.