[Haskell-cafe] (Un)termination of overloading resolution

[Ross Paterson]

On Tue, Feb 21, 2006 at 07:13:17PM -0800, oleg at pobox.com wrote:
> I'm afraid that may still be insufficient, as the following
> counter-example shows. It causes GHC 6.4.1 to loop in the typechecking
> phase. I haven't checked the latest GHC.

The HEAD is more cautious:

F.hs:12:0:
    Variable occurs more often in a constraint than in the instance head
      in the constraint: E m a b
    (Use -fallow-undecidable-instances to permit this)
    In the instance declaration for `Foo m (a -> ())'

This is required for all instances.  GHC 6.4 relaxed it in the presence
of FDs on the classes in the context, but as you've shown that is dangerous.