On Friday 22 October 2010 5:48:28 am Max Bolingbroke wrote:
I think evaluating dictionaries strictly is more of a "want to have" rather than "actually implemented". In particular, GHC supports building value-recursive dictionaries - and making dictionary arguments strict indiscriminately would make this feature rather useless.
It now occurs to me that I might be thinking of your paper on strict core. When does this come up? I only have one example that seems likely: data Mu f = In { out :: f (Mu f) } instance Show (f (Mu f)) => Show (Mu f) where show = show . out Is that an example of a value recursive dictionary? I also considered: data AnyShow = forall a. Show a => AS a instance Show AnyShow where show (AS x) = "AS (" ++ show x ++ ")" inf = AS inf but I don't think the dictionary is actually recursive there. Translating to explicit dictionary passing seems to confirm the latter impression. The former is more iffy. The most obvious way to build a value recursive dictionary would be an existential like: inf' = case inf' of AS x -> AS (x, x) but this is prevented from happening by the lack of irrefutable match.
What is definitely not OK is lazy pattern matching on GADTs which have *type equality* constraints on their existentials.
This is certainly the more compelling deal breaker. -- Dan