I'm more than open to switching to a more efficient solution.

-Edward

On Fri, Jul 17, 2015 at 5:16 PM, David Feuer <david.feuer@gmail.com> wrote:

The implementation could use some work. Something more like this, I think.

(<|>) = foldr (\x k ys -> x : k (drop 1 ys)) id

In pointless silliness, it's even

foldr ((. (. (maybe ([]) snd . uncons))) . (.) . (:)) id

On Jul 17, 2015 2:15 PM, "Edward Kmett" <ekmett@gmail.com> wrote:

Er.. I obviously meant

instance Alternative ZipList where

   empty = ZipList []

   ZipList xs <|> ZipList ys = ZipList (xs ++ drop (length xs) ys)

sorry for any confusion.

On Fri, Jul 17, 2015 at 11:37 AM, Edward Kmett <ekmett@gmail.com> wrote:

The ZipList data type admits one legal Alternative instance

instance Alternative ZipList where

   pure = ZipList []

   ZipList xs <|> ZipList ys = ZipList (xs ++ drop (length xs) ys)

The legality of this instance was first noted in http://people.cs.kuleuven.be/~tom.schrijvers/Research/papers/ppdp2015.pdf

I propose adding this instance, which acts like a generalized version of the Alternative for Maybe, choosing with a left bias.

Discussion Period: 2 weeks

-Edward

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

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