AssertionError: invalid prefix MethodType ... #16842
Comments
mateusz-lichota
added
itype:bug
itype:crash
stat:needs triage
|
Hi! Thank you for submitting the issue. Fortunately, this appears to be fixed on main (21729d2) and in 3.3.0-RC2, which means that this will be fixed in the next version of Scala (3.3.0). Instead of a crash we now get errors, all of which appear to be correct: -- [E007] Type Mismatch Error: main.scala:24:8 ---------------------------------
24 | Liter(SemanticArray[SemanticInt.type], x)
| ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
| Found: Int => SemanticArray[SemanticInt.type]
| Required: SemanticArray[SemanticType]
|
| longer explanation available when compiling with `-explain`
-- [E007] Type Mismatch Error: main.scala:24:41 --------------------------------
24 | Liter(SemanticArray[SemanticInt.type], x)
| ^
|Found: (x : List[Expr2[SemanticInt.type]])
|Required: ?1.T
|
|where: ?1 is an unknown value of type (dim: Int): SemanticArray[SemanticInt.type]
|
| longer explanation available when compiling with `-explain`As an extra, here is a version that fixes the above errors (and also compiles on 3.2.2): sealed trait Expr1
sealed trait Literal extends Expr1
case class ArrayLiter(elems: List[Expr1]) extends Literal
sealed trait SemanticType {
type T // the type with which a literal of this semanticType is represented
}
case object SemanticInt extends SemanticType {
type T = Int
}
case class SemanticArray[U <: SemanticType](dim: Int) extends SemanticType {
type T = List[U]
}
sealed trait Expr2[+T]
class Liter[T <: SemanticType](val ty: T, val value: ty.T) extends Expr2[T]
def typecheckArrayLiter(
a: ArrayLiter
): Liter[SemanticArray[? <: SemanticType]] = { // bit of a hack here, usually we would just make SemanticArray type parameter covariant, but here we would get an error: "covariant type U occurs in invariant position in type = List[U] of type T"
val x: List[SemanticInt.type] = List() // removed Expr2 from type declaration
Liter(SemanticArray[SemanticInt.type](0), x) // added a missing argument
}
|
jchyb
added
stat:cannot reproduce
and removed
stat:needs triage
|
I would expect "cannot reproduce" to mean the reported symptom can't be reproduced in a lab setting, rather than "can't be reproduced because the bug is already fixed." (I happened to arrive here because "cannot reproduce" is also a challenge.) |
|
@som-snytt you are right "cannot reproduce" should be used when we can't be reproduced in a lab setting.
It does fail as reported in 3.2.2. |
Compiler version
3.2.2
Minimized code
Output (click arrow to expand)
The text was updated successfully, but these errors were encountered: