Aside: If we decide that something like union subtypes are useful, it's worth noting that they seem to straddle the boundary between constraints and [overlapping] subtypes. I suspect that we *really* don't want to treat non-overlapping subtypes as subtypes in the inference engine (but I may be wrong). What I'm wondering is this:
Given a declared single-variable type class "tc 'a", should we permit "tc" to appear (syntactically) as a type name, where we understand x : tc to mean x : (forall tc 'a => 'a) Note: the intent of this is that 'a is a fresh type variable and the forall is understood to reside in the appropriate quantification scope in spite of appearing lexically in this position. On the one hand, this would sure reinforce a bunch of my initial confusions about type classes back when I was first getting used to them. On the other, it sure seems like a useful convenience syntax in an awful lot of situations. shap
_______________________________________________ bitc-dev mailing list bitc-dev@coyotos.org http://www.coyotos.org/mailman/listinfo/bitc-dev
