Clojurians Log v2
other-languages
Is anyone here by any chance doing the exercises from HaskellBook? I need a hint on assignment S from chapter 21
I don’t know much about that book but can you paste the problem in here? Or, go to #haskell on freenode…
@ddellacosta I need to write a Traversable instance for data S n a = S (n a) a deriving (Eq, Show). This is the code I have:
{-# LANGUAGE FlexibleContexts #-}
module TraversableS where
import Test.QuickCheck hiding (Success, Failure)
import Test.QuickCheck.Checkers
import Test.QuickCheck.Classes
import Data.Monoid
data S n a = S (n a) a deriving (Eq, Show)
instance ( Functor n
, Arbitrary (n a)
, Arbitrary a )
=> Arbitrary (S n a) where
arbitrary =
S <$> arbitrary <*> arbitrary
instance ( Applicative n
, Testable (n Property)
, EqProp a )
=> EqProp (S n a) where
(S x y) =-= (S p q) =
(property $ (=-=) <$> x <*> p)
.&. (y =-= q)
instance (Functor n) => Functor (S n) where
fmap f (S fa a) = S foo (f a) where
foo = (fmap f fa) -- we must do this to keep it of the type S (n b) b
instance (Foldable n) => Foldable (S n) where
foldMap f (S fa a) = foo <> f a where
foo = foldMap f fa
instance (Traversable n)
=> Traversable (S n) where
traverse f (S fa a) = s where
x = traverse f fa
y = f a
s = S <$> x <*> y
mainS = -- why you not work?
let trigger :: (S [] (Int, Int, [Int]))
trigger = undefined
in quickBatch (traversable trigger)
traversable:
fmap: *** Failed! Falsifiable (after 12 tests and 2 shrinks):
<function>
S [-11,8,-7,-9,-5,3,1,11,2,-10,6] (-8)
LHS
1
foldMap: +++ OK, passed 500 tests.
Some of the code was already given in the book, but not the Foldable and Traversable instances
just include DeriveFoldable and DeriveTraversable? 😛
lol
seriously, not sure exactly what it’s testing so not sure I can help, but let me stare at it a bit
I’m afk for 15 minutes
and maybe someone else who is better at Haskell will show up too
@borkdude so is S [-11,8,-7,-9,-5,3,1,11,2,-10,6] (-8) an instance of S n a, where a is Int and n is [] (a list)?
if I’m reading the types correctly, it seems like the foldMap instance is treating the semigroup for n a and a as the same thing
but if it’s type-checking that suggests I’m confused
@ddellacosta that’s correct, n is [] and a is Int
@ddellacosta hmm, it does type-check, but let me check this
it also law-checks
no I’m sorry, I was confused
so yeah your Foldable seems pretty alright
yeah, foo is of type m and so is f a
so what about the nevermind, was misreadingtraverse then…are you sure you want an S n a return value? like if you have a (S n a) and n is Applicative, maybe that needs to be n (S n b) (where b is the output of your a -> f b)?
but again, if it’s type-checking I may be confused
I’m just comparing your code to the type sig for traverse
…and hoping my questions maybe trigger you realizing something, even if I’m offbase
the type here is (a -> f b) -> S n a -> f (S n b)
so for example:
-- example = S (Just "http://...") "http://..."
-- get :: String -> IO Resp
-- fmap get example :: S (Just (IO resp)) (IO resp)
-- but we want: IO (S (Just resp) resp)
this is very confusing
(thinking/reading)
yeah this is a tough one, lots of moving parts
shoot @borkdude I have to go, I’m sorry, if I can get back on and help more I will--you may have more luck posting this in #haskell on IRC though, I bet with the level of folks there someone will immediately point out the issue
Example:
type Resp = String
get :: String -> IO Resp
get s = pure "<html></html>"
example = S (Just "http://") "http://"
test1 :: S Maybe (IO Resp)
test1 = fmap get example
test2 :: IO (S Maybe Resp)
test2 = traverse get example -- S (Just "<html></html>") "<html></html>"
Thanks!
wait so with that example, that’s showing you how it does work now? (except the return value of the last one is in IO I guess, assuming you dumped that out in ghci)
yeah it does, just tried it, okay
damn I can’t see what’s wrong with it, for example just tried this too:
λ> traverse (\n -> [n+1]) (S [1,2,3] 1)
[S [2,3,4] 2]
λ>
okay actually going away now
thanks. The full code is here in this gist. I’ll start asking others as well: https://gist.github.com/borkdude/3d7eebe5fd2f7ac1fdca73a8cceb608e
@ddellacosta mystery solved, the EqProp instance from the book was wrong