Regression in match exhaustivity of higher-kinded types

[WojciechMazur]

Compiler version

3.3.1-RC1-bin-20230407-bcaa1ca-NIGHTLY

Minimized code

Based on Open CB failure https://github.com/VirtusLab/community-build3/actions/runs/4650721329/jobs/8230332752 for rescala-lang/rescala

import scala.util.* 

trait Transaction {
  type State[_]
}
object Transaction {
  type of[S[_]] = Transaction { type State[A] = S[A] }
}
trait DynamicScope[State[_]]

case class ScopeSearch[State[_]](self: Either[Transaction.of[State], DynamicScope[State]]) {

  def embedTransaction[T](f: Transaction.of[State] => T): T =
    self match {
      case Left(integrated) => ??? 
      case Right(ds)        => ???
    }
}

Output

Compiling project (Scala 3.3.1-RC1-bin-20230407-bcaa1ca-NIGHTLY, JVM)
[warn] ./test.scala:13:5
[warn] match may not be exhaustive.
[warn] 
[warn] It would fail on pattern case: Left(_)
[warn]     self match {
[warn]     ^^^^
Compiled project (Scala 3.3.1-RC1-bin-20230407-bcaa1ca-NIGHTLY, JVM)

Expectation

Should not give warning

[WojciechMazur]

Bisect points to a range of commits 0df5ae2 .. 8020c77 , but further bisect was failing due to some other issues:

Last good release: 3.3.1-RC1-bin-20230301-0df5ae2-NIGHTLY
First bad release: 3.3.1-RC1-bin-20230302-8020c77-NIGHTLY

[scala-center-bot]

This issue was picked for the Issue Spree No. 29 of 18 April 2023 which takes place in a week from now. @dwijnand, @natsukagami, @TheElectronWill, @XLPI will be working on it. If you have any insight into the issue or guidance on how to fix it, please leave it here.

[dwijnand]

Minimised

trait Foo:
  type Bar[_]

object Foo:
  type Aux[B[_]] = Foo { type Bar[A] = B[A] }

class Test:
  def test[B[_]](self: Option[Foo.Aux[B]]) = self match
    case Some(b)             => 1
    case None                => 2

[XLPI]

Guys, is anybody continue to work on this issue? How actually our workflow in Compiler Academy is should going? @scala-center-bot

[scala-center-bot]

This issue was picked for the Issue Spree No. 30 of 09 May 2023 which takes place in a week from now. @dwijnand, @XLPI will be working on it. If you have any insight into the issue or guidance on how to fix it, please leave it here.

[mbovel]

Guys, is anybody continue to work on this issue? How actually our workflow in Compiler Academy is should going?

I re-assigned you to work on this with Dale during next spree. Is that okay with you?

[mbovel]

Please ignore the 2 previous message from the Scala Center bot. Teams for the next spree can be found here: https://airtable.com/shrSni1OItSuErosU/tblLEWc2ISeGHPbsc. Sorry for the spam!

[mbovel]

What is the current status of this issue? It has been investigated during 2 sprees; what were the conclusions?

[dwijnand]

If I remember correctly, it's due to a difference in how type variables are instantiated in the Space engine. I'll got this on my list to come back to.