Using Arrows for Dependency Handling

Author of the article here. I'll try to give a quick informal explanation. If you look at the type signature of `>>=`:

  m a -> (a -> m b) -> m b

Let's suppose there's a HakyllAction monad. The >>= signature would be:

  HakyllAction a -> (a -> HakyllAction b) -> HakyllAction b

If we implement this operator, we get a value `HakyllAction a` and a function `a -> HakyllAction b`. We can look at the dependencies of the value, but we don't have access to the dependencies of the `a -> HakyllAction b`. If we want that, we need to apply it to the first value.

This goes against the idea of dependency handling -- we want to know the dependencies first, and then only execute the function if needed.

Hope this helps.