Rewrite implicit as given? implicitly as summon?

[gvolpe]

Does a rule like this exist? Scala 3 provides the -new-syntax, -indent and -rewrite flags, but this doesn't seem supported. Wonder if this is something for scalafix instead?

I would love to have a configuration like the following:

runner.dialect = scala3
style = defaultWithAlign
rewrite {
  scala3 {
    convertToNewSyntax = true
    removeOptionalBraces = true
    convertImplicitsToNewSyntax = true
  }
  trailingCommas.style = never
}

Where given the following Scala code:

final case class Person(age: Int)

implicit val showPerson: Show[Person] = Show.fromToString

trait Log[F[_]] {
   def info(msg: => String): F[Unit]
}

object Log {
  def apply[F[_]: Log]: Log[F] = implicitly

  object NoOp {
    implicit def instance[F[_]: Applicative]: Log[F] =
      new Log[F] {
        def info(msg: => String): F[Unit] = Applicative[F].unit
      }
  }
}

trait Log2[F[_]] {
   def info(msg: => String): F[Unit]
   def error(msg: => String): F[Unit]
}

object Log2 {
  def apply[F[_]: Log2]: Log2[F] = implicitly

  object NoOp {
    implicit def instance[F[_]: Applicative]: Log2[F] =
      new Log2[F] {
        def info(msg: => String): F[Unit] = Applicative[F].unit
        def error(msg: => String): F[Unit] = Applicative[F].unit
      }
  }
}

It would produce the following formatted code:

final case class Person(age: Int)

given Show[Person] = Show.fromToString

trait Log[F[_]]:
   def info(msg: => String): F[Unit]

object Log:
  def apply[F[_]: Log]: Log[F] = summon

  object NoOp:
    given [F[_]: Applicative]: Log[F] with
      def info(msg: => String): F[Unit] = Applicative[F].unit

trait Log2[F[_]]:
   def info(msg: => String): F[Unit]
   def error(msg: => String): F[Unit]

object Log2:
  def apply[F[_]: Log2]: Log2[F] = summon

  object NoOp:
    given [F[_]: Applicative]: Log2[F] = new:
      def info(msg: => String): F[Unit] = Applicative[F].unit
      def error(msg: => String): F[Unit] = Applicative[F].unit

[kitbellew]

no, it doesn't exist.

[gvolpe]

@kitbellew do you think this should be supported by scalac first? or do you see this as a potential feature for scalafmt? Thanks!

[tgodzik]

This should probably be a warning with actionable diagnostic coming from the compiler. We could then apply i automatically.