I'm thinking something like these should be in the standard libraries in some form. The first (fromReal) is pretty straightforward. The others come from the observation that there's no reason why the arguments to div and mod should be integers. The basic formulae are these: a `div` b = floor (a/b) a `mod` b = a - (a div b) * b As you can see, div should return an integer (because that's what floor does), but everything else generalises to real. I suspect my implementations are not optimal. -- | similar idea to "fromIntegral" fromReal :: (Real a,Fractional b) => a -> b fromReal = fromRational . toRational -- | like "div", but with a more useful type div' :: (Real a,Integral b) => a -> a -> b div' n d = floor ((toRational n) / (toRational d)) -- | like "divMod", but with a more useful type divMod' :: (Real a,Integral b) => a -> a -> (b,a) divMod' n d = (f,n - (fromIntegral f) * d) where f = div' n d -- | like "mod", but with a more useful type mod' :: (Real a) => a -> a -> a mod' n d = n - (fromInteger f) * d where f = div' n d -- Ashley Yakeley, Seattle WA