'case' with 'given' error has precedence over unreachable case

[sjbiaga]

I've run into this weird behavior by chance, between "old" code I had and its refactoring.

Compiler version

3.6.2-RC3

Minimized example

First, the "old" code was pattern matching an Option[(List[Int], Map[Int, Double])], but I dropped the Option, leaving just the (pair) tuple. Yet, Some in pattern matching remained:

object X:
  def m(using List[Int])(using Map[Int, Double]) = {}
  def apply(using t2: (List[Int], Map[Int, Double])) =
  t2 match
    case Some((given List[Int], given Map[Int, Double])) => m
    case _ =>

So I got this:

Output Error message

5 |      case Some((given List[Int], given Map[Int, Double])) => m
  |                                                               ^
  |No given instance of type List[Int] was found for parameter x$1 of method m in object X
  |
  |Note: given instance given_List_Int was not considered because it was not imported with `import given`.
1 error found

Why this Error was not helpful

Second, if I now removed the call to method m...

object X:
  def m(using List[Int])(using Map[Int, Double]) = {}
  def apply(using t2: (List[Int], Map[Int, Double])) =
  t2 match
    case Some((given List[Int], given Map[Int, Double])) =>
    case _ =>

.. the "correct" error is now:

-- [E030] Match case Unreachable Warning: --------------------------------------
5 |      case Some((given List[Int], given Map[Int, Double])) =>
  |           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  |           Unreachable case
-- Error: ----------------------------------------------------------------------
5 |      case Some((given List[Int], given Map[Int, Double])) =>
  |           ^
  |this case is unreachable since type (List[Int], Map[Int, Double]) is not a subclass of class Some
1 warning found
1 error found

The first message was misleading because it actually discovered the Some-wrapped given List[Int], whereas by the second message, the case was unreachable.

I know that Scala compiler reports from "right to left", so I'm not exactly sure.

[sjbiaga]

Already fixed in 3.6.3-RC1.