Despite the theoretical decidability of rank-2 type inference their actual implementations in GHC are the same IIRC (so rank-2 types are not fully inferrable by

Despite the theoretical decidability of rank-2 type inference their actual implementations in GHC are the same IIRC (so rank-2 types are not fully inferrable by GHC). I think the difference is really just one of historicity and backwards compatibility at this point. They resolve to the same underlying mechanism.