Tony Morris

@nuttycom @jedws @puffnfresh Not winning, you just yelled at us for continuing the argument. :-)

@djspiewak @nuttycom @puffnfresh ha! interesting q. though, as it depends on maintaining stable equality, so it is but with qualifications

@jedws @nuttycom @puffnfresh If you do care about efficiency and you don’t have a universal hashCode/equals, then you need Kris’s Functorial

@djspiewak @jedws @nuttycom @puffnfresh equals/hashCode is irrelevant; parametricity says you can't look at the elements, period

@tpolecat @jedws @nuttycom @puffnfresh JVM gives us the following: equals :: forall a . a -> a -> Boolean hashCode :: forall a . a -> Int

@djspiewak @tpolecat @jedws @nuttycom no, it doesn't.

@puffnfresh @tpolecat @jedws @nuttycom You can argue about whether or not that’s a good thing (it’s not, imo), but it is a thing.

@djspiewak @puffnfresh @tpolecat @nuttycom right, definitely not, for various reasons, but Functor[Set] breaks the composition law