If you type :i Monad in ghci, you will see this instance

instance Monad ((->) r)

What this means is a function where the first argument is of type 'r'... is a monad.  You can in fact use do notation / return on a tuple to manipulate its second argument monadically.

So let's look at what that does to the type signature of join when 'm' is ((->) b)

join :: Monad m => m (m a) -> m a

 -- m = ((->) b)

join :: ((->) b ((->) b a)) -> (((->) b a))

Now we just have to move the arrows from prefix to infix.  Let's do it step by step.

join :: ((->) b (b -> a)) -> (b -> a)
join :: (b -> (b -> a)) -> (b -> a)

x -> (y -> z) is equivalent to x -> y -> z

join :: (b -> b -> a) -> (b -> a)
join :: (b -> b -> a) -> b -> a

So now when you put an operator into it that takes two arguments

(,) :: a -> b -> (a,b)

You get the type you saw.

join (,) :: b -> (b, b)





On Fri, Dec 8, 2017 at 10:37 AM, Quentin Liu <quentin.liu.0415@gmail.com> wrote:
Hi,

The function `join` flattens a double-layered monad into one layer and its type signature is 

  join :: (Monad m) => m (m a) -> m a

But when the first argument supplied is `(,)`, the type signature becomes 

  join (,) :: b -> (b, b)

in ghci. The monad constraint is lost when supplied the first argument. So my question is why the type constraint is lost and what monad is supplied here.

Regards,
Qingbo Liu

_______________________________________________
Beginners mailing list
Beginners@haskell.org
http://mail.haskell.org/cgi-bin/mailman/listinfo/beginners