Re: [Haskell-cafe] the last mile in haskell performance