6.10 regression: panic in tyFamInsts on invalid code

{-# LANGUAGE ScopedTypeVariables, Rank2Types #-}
module Asdf where

g :: Num a => a -> (forall a. a -> Int) -> Int
g = flip id

f = g 0 $ \(_ :: forall a. Num a => a) -> 0

>ghc -c asdf.hs
ghc: panic! (the 'impossible' happened)
  (GHC version 6.10.0.20081002 for i386-unknown-mingw32):
        types\Type.lhs:(881,0)-(889,46): Non-exhaustive patterns in function tyFamInsts

With 6.8.2 I get instead the following, which I think is correct:

] ghc -XPatternSignatures -c asdf.hs
asdf.hs:7:12:
    Couldn't match expected type `a'
           against inferred type `forall a1. (Num a1) => a1'
      `a' is a rigid type variable bound by
          the polymorphic type `forall a. a -> Int' at asdf.hs:7:4
    In the pattern: _ :: forall a. (Num a) => a
    In the second argument of `($)', namely
        `\ (_ :: forall a. (Num a) => a) -> 0'
    In the expression: g 0 $ \ (_ :: forall a. (Num a) => a) -> 0

On the other hand, replacing f with:

f = g 0 $ ((\_ -> 0) :: forall a. Num a => a -> Int)

Works fine. I don't know why the original code is invalid but it certainly shouldn't cause a panic.

1. 10 wasn't available in the "Version" listing so I filed this for 6.9.

Trac metadata

Trac field	Value
Version	6.9
Type	Bug
TypeOfFailure	OtherFailure
Priority	normal
Resolution	Unresolved
Component	Compiler
Test case
Differential revisions
BlockedBy
Related
Blocking
CC
Operating system
Architecture