Jumbled error message from type family operator

[lilac]

The attached file produces this error message:

Testcase.hs:82:0:
    Couldn't match expected type `True'
           against inferred type `Main.R::<=:GHC6'10GHC6'8 Minor1 minor'
    When generalising the type(s) for `testcase'

That should say

[...] against inferred type `GHC6'10 Minor1 :<=: GHC6'8 minor'

Apologies for the complexity of the testcase; simpler testcases seem to not reproduce the problem.

[lilac]

Testcase reproducing bad error message

[simonpj]

I agree. Thank you for the test case. Will look at it when the new type inference machinery is done. Weeks not months I hope.

Simon

[simonpj]

This bug is fixed by the latest wave of typechecker fixes (I'm not sure exactly which one). Now we get these errors:

T4174.hs:42:12:
    Couldn't match type `False' with `True'
    Expected type: True
      Actual type: GHC6'10 Minor1 :<=: GHCVersion (WayOf m)
    In the expression: sync_large_objects
    In an equation for `testcase': testcase = sync_large_objects

T4174.hs:42:12:
    Could not deduce (a ~ SmStep) from the context (Monad m)
      `a' is a rigid type variable bound by
          the type signature for `testcase' at T4174.hs:41:60
    In the expression: sync_large_objects
    In an equation for `testcase': testcase = sync_large_objects

T4174.hs:42:12:
    Could not deduce (b ~ RtsSpinLock) from the context (Monad m)
      `b' is a rigid type variable bound by
          the type signature for `testcase' at T4174.hs:41:62
    In the expression: sync_large_objects
    In an equation for `testcase': testcase = sync_large_objects

T4174.hs:42:12:
    Could not deduce (p ~ True) from the context (Monad m)
      `p' is a rigid type variable bound by
          the type signature for `testcase' at T4174.hs:41:57
    Expected type: True
      Actual type: Threaded (WayOf m)
    In the expression: sync_large_objects
    In an equation for `testcase': testcase = sync_large_objects

The first of these is the one you were after, and looks quite reasonable.

I've added a regression test.

Thanks for reporting it!

Simon