re free type variables in SML

[Stephen Weeks]

> In addition to the ambiguous of the definition with regards to top-level free
> type variables in the result of expansive expressions, it would be really great
> if the scope over which type inference could run was spelled out, right?

With the exception of overloading, I think the scope is spelled out.  Although I
think even easier is my proposal (soon to appear on comp.lang.ml) of not
allowing expansive declarations with unconstrained types.