Re: Alternative instance for Const

https://wiki.haskell.org/Typeclassopedia#Laws_6 On 21.3.2020 23.46, Simon Jakobi via Libraries wrote:

Hi Zemyla,

Could you clarify where the requirement for distributivity comes from?

The haddocks for Alternative don't mention it: http://downloads.haskell.org/~ghc/8.10.1-rc1/docs/html/libraries/base-4.14.0...

Thanks, Simon

Am Sa., 21. März 2020 um 21:23 Uhr schrieb Zemyla :

...

No, the Alternative instance should be based on semirings instead. The one you propose isn't distributive.

Lacking general semirings, the best we can do at the moment is newtypes for each semiring we have. For instance:

newtype NumConst a b = NumConst { getNumConst :: a } deriving (Functor)

instance Num a => Applicative (NumConst a) where pure = const (NumConst 1) (<*>) = (coerce :: (a -> a -> a) -> NumConst a (u -> v) -> NumConst a u -> NumConst a v) (*)

instance Num a => Alternative (NumConst a) where empty = NumConst 0 (<|>) = (coerce :: (a -> a -> a) -> NumConst a b -> NumConst a b -> NumConst a b) (+)

On Sat, Mar 21, 2020, 13:44 chessai . wrote:

...

We already have

instance Monoid m => Applicative (Const m)

we could easily add

instance Monoid m => Alternative (Const m) where empty = coerce mempty (<|>) = coerce (<>)

which trivially satisfies left/right identity and associativity.

I propose we add this instance to base. _______________________________________________ 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

---

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