[Haskell-cafe] Why does disambiguation in RebindableSyntax sometimes requires type signatures?

[Henning Thielemann]

On Mon, 2 Nov 2015, Jan Bracker wrote:

> Hi Henning,
> but with type signatures the errors do not occure.
> 
> My best guess for that one is that inferring the correct kinds for lifted data types is not done properly.

I have not followed the previous conversation. A general problem is that 
if a top-level function misses a type signature then GHC might choose a 
type that is too special, such that it fits one call site but mismatches 
another one. It is always a good idea to add type-signatures to top-level 
functions. If your problem occurs with local functions and you can solve 
it with a type signature then I think this is the way to go.