On 02/26/2016 12:41 PM, Rein Henrichs wrote:
Pointfree is good for reasoning about *composition*. It can often be more readable than pointful code when the focus of the function is on composition of other functions. For example, take this function from Bird's /Pearls of Functional Algorithm Design/:
boxes = map ungroup . ungroup . map cols . group . map group
Compare the pointful version:
boxes matrix = map ungroup (ungroup (map cols (group (map group matrix))))
Readibility is subjective, but I think many people will agree that the pointfree version is easier to read.
Sure, but does anyone have any idea what that first version is supposed to do? It would be much better to write it out: boxes matrix = one_big_list where -- The function (group . map group) produces a list of box -- matrices, transposing each one in the process. transposed_box_matrices = group (map group matrix) -- The "cols" function performs a transpose, so mapping it over -- transposed_matrices gives us the box matrices we want. box_matrices = map cols transposed_box_matrices -- Now we concat everything into one big matrix for some reason. -- This is the inverse of the (group . map group) that we did -- earlier. one_big_list = map ungroup (ungroup box_matrices) You don't even need to write a book to explain what it does =)