Yichen Xu

test performance please

test performance please

Hey Alex! @abgruszecki I have finished one round of PR polishing. I have done the following things: - I have refactored the interface of path aliasing constraints. It is now...

Ah, so this simplified snippet reveals a problem different from the original one in #15731 (since the list in that example is church-encoded and no selection is involved).

I tweaked the snippet a bit (defining the wrapper in church-style) and found this (which is closer to the original list encoding example and seems to reveal the same problem):...

Maybe we should issue an error when applying `strictMap[{*} Unit -> A, A](mx)` on `force` (on the last line). The partially-applied function `strictMap[{*} Unit -> A, A](mx)` expects an argument...

Note that there are cases where we do have bounds as GADT constraints: ```scala enum C[-T]: case C1() extends C[EmptyTuple] def foo[A](e: C[A]) = e match { case C.C1() =>...

More explanation: after having a closer inspection of the issue, the call site mentioned in the original comment is incorrect and misleading. What actually happens: - We are trying to...

To verify, I added the tracing expression `trace.force(i"$tp1 . $name", show = true) { tp1.member(name) }`. This is what it prints: ``` ==> X . T?

Similarly, we will wait for a refactorization of GADT (https://github.com/lampepfl/dotty/pull/15928) before proceeding this feature. I will close this for now and go back to it after the refactorization is done.