Fix #16405 ctd - wildcards prematurely resolving to Nothing #16764
Conversation
This was a problem because it could it get in the way of some metaprogramming techniques. The main issue was the fact that when typing functions, the type inference would first look at the types from the source method (resolving type wildcards to Nothing) and only after that, it could look at the target method. Now, in the case of wildcards we save that fact for later (while still resolving the prototype parameter to Nothing) and we in that case we prioritize according to the target method, after which we fallback to the default procedure.
|
@jchyb It turned out that we don't need to detect wildcards arguments. It's simpler and equally valid to just avoid |
|
|
||
| object O { | ||
| def m(i: Int): Int = i | ||
| val x: Q[_] = m // error: result type of lambda is an underspecified SAM type Q[?] |
There was a problem hiding this comment.
Previously I was worried too much that this approach as a whole would create some unreachable code in typedClosure:
def typedClosure(tree: untpd.Closure, pt: Type)(using Context): Tree = {
val env1 = tree.env mapconserve (typed(_))
val meth1 = typedUnadapted(tree.meth)
val target =
if (tree.tpt.isEmpty)
meth1.tpe.widen match {
case mt: MethodType =>
pt.findFunctionType match {
case pt @ SAMType(sam)
if !defn.isFunctionType(pt) && mt <:< sam =>
// SAMs of the form C[?] where C is a class cannot be conversion targets.
// The resulting class `class $anon extends C[?] {...}` would be illegal,
// since type arguments to `C`'s super constructor cannot be constructed.
def isWildcardClassSAM =
!pt.classSymbol.is(Trait) && pt.argInfos.exists(_.isInstanceOf[TypeBounds])
val targetTpe =
if isFullyDefined(pt, ForceDegree.all) && !isWildcardClassSAM then
pt
else if pt.isRef(defn.PartialFunctionClass) then
// Replace the underspecified expected type by one based on the closure method type
defn.PartialFunctionOf(mt.firstParamTypes.head, mt.resultType)
else // Unreachable
report.error(em"result type of lambda is an underspecified SAM type $pt", tree.srcPos) //
pt
TypeTree(targetTpe)But I guess it is should not be a problem, since the checks in the above make the code future proof.
|
Yes, there's a problem with ZIO. I'll try to find out more what caused it. |
|
@jchyb It turned out that one needs to take wildcards into account to make both zio and #16405 pass. Making type inference do the expected thing in all the edge cases is really a minefield. My latest changes essentially come back to your idea, but instead of keeping track of wildcards in a separate data structure, we simply keep the original wildcard bounds in decomposeProto. |
3 participants
Fixes #16405, which was a problem because it could it get in the way of some metaprogramming techniques. The main issue was the fact that when typing functions, the type inference would first look at the types from the source method (in decomposeProtoFunction resolving found type wildcards to Nothing) and only after that, it could look at the target method.
This is a continuation and simplification of #16625