Should PartialOrd be a superclass of joins and meets?

[ocharles]

If "n mathematics, a join-semilattice (or upper semilattice) is a partially ordered set that has a join (a least upper bound)", shouldn't we have

class PartialOrd a => JoinSemiLattice a where
class PartialOrd a => MeetSemiLattice a where

?

This would have pre-empted the problems reported in #53 and #52.

[phadej]

That will make life hard for types without decidable equality, like JoinSemiLattice v => JoinSemiLattice (k -> v)

[ocharles]

Ah, that is problematic.