Haskell map/zip Vs. list comprehension

Question

Which of the following are you most likely to write?

r = zip xs $ map sqrt xs

or

r = [(x, sqrt x) | x <- xs]

Sample code on the Internet seems to indicate that the former is more abundant and the preferred way.

Answer 1

People who spend too much time in #haskell would probably write that as

r = map (id &&& sqrt) xs

(&&&) is a fun combinator defined in Control.Arrow. Its actual type signature is complicated because it's generalized to all instances of Arrow. But it's often used with the (->) instance of Arrow, which results in this type signature:

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

Answer 2

Although I tend to not use them very often, in this case, I think I'd prefer the list comprehension version, since it seems cleaner to me.

If you're into point free style, you might like this one, too:

f = zip `ap` map sqrt

ap lives in Control.Monad and in this case, it can be thought of as the S combinator, which generalizes application in SKI calculus:

ap f g x == f x (g x)
ap const const == id

As Conal points out, this may also be generalized from Monad to Applicative thusly (import Control.Applicative):

f = zip <*> map sqrt

Answer 3

I would probably write map/zip and then later wish I had written the list comprehension.