On Wed, Apr 1, 2015 at 9:08 AM, Jonathan S. Shapiro <[email protected]> wrote: > On Wed, Apr 1, 2015 at 4:56 AM, William ML Leslie > <[email protected]> wrote: >> Unification is really the wrong word to describe what we are doing during >> arity specialisation. Function types of concrete arity only unify with >> themselves. Arity-abstract function types also unify only with themselves. >> They may be specialised, in which case a substitution is performed, but this >> substitution is not term for term but term for a specific variable. > > > Yes.
I don't understand. Can't we unify (cfn 'a 'b->'c) with (afn 'a 'b->'c) to get (cfn 'a 'b->'c)? Why is this not a unification? I don't understand any of this talk about it being outside HM inference. Wasn't that true as soon as we had type-level variables that aren't types? _______________________________________________ bitc-dev mailing list [email protected] http://www.coyotos.org/mailman/listinfo/bitc-dev
