• removed Debug.Foreign
• added comment about matches

Don't sure that I understand why there is shouldn't be two Eff in matches. Second argument is effectful callback, that called after hash changed and can be matched with first argument. If previous hash can be matched too, then first argument of callback is Just a else -- Nothing

The point is that the type signature is extremely specific and can be made more generic. Generic code is easier to get right (there are fewer ways it can go wrong) and is useful solving strictly more problems. The only reason matches / matches' wants to be specific is so it can create an "empty", so there are really two possibilities to make it more generic: require an empty (Monoid) or push that off on the user (return a Maybe).

matches1 :: forall e a b. (Monoid b) => Match a -> (Maybe a -> a -> b) -> b

matches2 :: forall e a b. Match a -> (Maybe a -> a -> b) -> Maybe b

Look at how the old one can be implemented in terms of either of these:

matches3 m f = matches1 m f -- assuming monoid for Eff eff Unit

matches4 m f = fromMaybe (pure unit) (matches2 m f )

In addition to these points, pure code shouldn't require effects when it doesn't have to. It's very surprising to see a matching function expect an effectful callback, it implies the matcher function is going to do something effectful, too, which is not the case.