Re: Overloaded unification

On Jan 27, 10:35=A0pm, "David B. Benson" <dben...@[EMAIL PROTECTED]
> wrote:
> On Jan 26, 2:58 pm, Jon Harrop <use...@[EMAIL PROTECTED]
> wrote:
>
> > ... Statically-resolved overloading basically
> > completely undermines type inference ...
>
> I don't think this is true. =A0What happens is that the type
> inference algorithm becomes worse-case exponential time.
> However, if there is only a little bit of overloading, modern
> machines are so fast one won't notice the extra time taken to
> resolve the overloads.

HM type inference already has worst-case exponential complexity. The
bigger problem with ad-hoc overloading is that you loose principal
types, which pretty much breaks the algorithm.

> Like any tool, this one can be misused. <i>caveat user!</i>

Right, the more interesting question is: can it be used for anything
good?

- Andreas