Automate migration to 1.0 with Scalafix

[gabro]

Now that the 1.0 is close, I got a chance to take a look at the breaking changes it introduces and I noticed that the vast majority of them are well within the possibilities of Scalafix.

So, well, why not just automate the migration away?
I started experimenting on it and it's promising! So far I got:

• removal of Unapply machinery (just a bunch of symbol renames, but, hey, I wanted to start easy)
• removal of CartesianBuilder syntax (this is trickier, but the rewrite is still around 50 lines of code)

Would you be interested in this?

If so, I think it would be really cool to ship the rewrites with the final release of 1.0.0 so that anyone can access them easily simply by having cats as a dependency. I'll push my work in progress soon, but in the meantime here's some diff I was able to fully automate:

   import cats.instances.option._
-  import cats.syntax.cartesian._
+  import cats.syntax.apply._
   val o1: Option[Int] = Some(42)
   val o2: Option[String] = Some("hello")
   val o3: Option[Int] = Some(2)
-  (o1 |@| o2).map((i: Int, s: String) => i.toString ++ s)
-  (o1 |@| o2).tupled
-  (o1 |@| o2 |@| o3).map(_ + _ + _)
+  (o1, o2).mapN((i: Int, s: String) => i.toString ++ s)
+  (o1, o2).tupled
+  (o1, o2, o3).mapN(_ + _ + _)

   import cats.{Semigroup, Eq}
   import cats.instances.string._
   import cats.instances.double._
   case class Foo(a: String, c: List[Double])

-  (Semigroup[String] |@| Semigroup[List[Double]]).imap(Foo.apply)(Function.unlift(Foo.unapply))
-
-  (Eq[Double] |@| Eq[String]).contramap { (a: Foo) => (2, "bar") }
+  (Semigroup[String], Semigroup[List[Double]]).imapN(Foo.apply)(Function.unlift(Foo.unapply))
+  (Eq[Double], Eq[String]).contramapN { (a: Foo) => (2, "bar") }

A couple of more things to notice:

• the rewrites are semantic, meaning they're using the semantic information not just the syntactic ones (e.g. when renaming map we're looking for _root_.cats.syntax.CartesianBuilder#CartesianBuilder2.map., not just any map)
• I'm currently targeting the 0.9.0 -> 1.0.0 migration, although this would probably be easy to extend to prior versions

[kailuowang]

Love the idea. It also makes it easier for people to start contribute to #1762

[gabro]

#1793 is now merged!

There are still a lot of low-hanging fruits, but in the meantime it would be great if anyone could try running the rewrites in repos using cats and report feedbacks here.

Instructions here https://github.com/typelevel/cats/blob/master/scalafix/README.md

[kailuowang]

Thanks so much again. I am waiting on this PR #1806 before updating the release note. Then I am thinking we should broadcasting it on twitter, do you mind if I mention your twitter handler?

[gabro]

happy to finally contribute back! :) I tweeted it out myself, but I guess some more noise won't hurt ;)

https://twitter.com/gabro27/status/895289070628294656

[mfirry]

I've just tried this migration and I get a weird import fix:

-import cats.syntax.cartesian._
+import cats.syntax.semigroupal._import cats.syntax.apply._

[ceedubs]

This was mostly resolved by #1793, and at this point Cats 1.0 has been out long enough that I doubt anyone is only now upgrading to it. I'm going to go ahead and close this out.