On: Beating Decades of Optimized C with 80 Lines of Haskell

[olliej]

You are correct - I was wrong about utf8, it is just doing a more correct decode than the inline trivial test that the Haskell version does.