On Wed, Feb 18, 2015 at 1:53 PM, Matt Oliveri <[email protected]> wrote:
> Why can't there be a mutable slot of an arity-abstract type? I mean, > implementation-wise, I see why it cannot exist, but what is it in the > proposal that prevents it? (Guess: Arity-abstract types are actually > constraints on quantified arity-concrete types, and BitC won't support > quantifiers under other type constructors. Or even if it does, there's > no way to make a function with multiple native arities.) > I over-stated the restriction. Mutable slots are monomorphic, so technically there *can* be a mutable slot of arity-abstract type. But as you all have figured out by now, my brain tends to thing about concrete stuff first. By the time you run the first instruction of main(), all of those type variables need to be specialized. Thus my original (not quite correct) statement. That said, two different instantiations of a type can certainly specialize a mutable (therefore monomorphic) slot in different ways. shap
_______________________________________________ bitc-dev mailing list [email protected] http://www.coyotos.org/mailman/listinfo/bitc-dev
