> a)). Mycroft first suggested a type system that would allow such
> functions to be typed, but I think it is still an open question as to
> whether an inference algorithm exists for the type system. (There was
> a paper published that claimed an algorithm, but it was later withdrawn
> as incorrect.)
I heard that it was proved impossible when semiunification turned out
to be undecidable. (But then again this not my area, so I might be wrong.)
But the problem is easily solved in practice, because you can do
typechecking instead of type inference. All you have to do is to
give type signatures for the problematic functions. Currently this
is not allowed in Haskell, but I think this is what SML and Miranda
does. (Also hbc if you don't use pedantic.)
-- Lennart
