That's a good idea, though not understanding the state monad is still pretty frustrating. On 2/12/16 6:38 AM, Nikita Kartashov wrote:

Hi!

Take a look at MonadRandom [1]. It is basically what you want without getting generator explicitly.

[1] https://hackage.haskell.org/package/MonadRandom

With regards, Nikita Kartashov

On 12 Feb 2016, at 04:14, Thomas Jakway mailto:tjakway@nyu.edu> wrote:

...

I'm having a bad time using the State monad to generate random numbers without carrying around a lot of StdGens manually. I have this snippet in the IO monad:

... IO stuff ... gen <- getStdGen let (numPlayers, numMatches) = (evalState genRandVariables gen) :: (Integer, Integer) ... More IO stuff ...

where maxRandPlayers = 10 :: Integer minRandMatches = 10 :: Integer maxRandMatches = 100 :: Integer genRandVariables = (do np <- randomR (1, maxRandPlayers) --minimum 1 other player nm <- randomR (minRandMatches, maxRandMatches) return (np, nm)) :: State StdGen (Integer, Integer)

I get this error message: test/Jakway/Blackjack/Tests/IntegrationTests/MatchTests.hs:53:23: Couldn't match expected type ‘StateT StdGen Data.Functor.Identity.Identity Integer’ with actual type ‘g0 -> (Integer, g0)’ Probable cause: ‘randomR’ is applied to too few arguments In a stmt of a 'do' block: np <- randomR (1, maxRandPlayers) In the expression: (do { np <- randomR (1, maxRandPlayers); nm <- randomR (minRandMatches, maxRandMatches); return (np, nm) }) :: State StdGen (Integer, Integer)

test/Jakway/Blackjack/Tests/IntegrationTests/MatchTests.hs:54:23: Couldn't match expected type ‘StateT StdGen Data.Functor.Identity.Identity Integer’ with actual type ‘g1 -> (Integer, g1)’ Probable cause: ‘randomR’ is applied to too few arguments In a stmt of a 'do' block: nm <- randomR (minRandMatches, maxRandMatches) In the expression: (do { np <- randomR (1, maxRandPlayers); nm <- randomR (minRandMatches, maxRandMatches); return (np, nm) }) :: State StdGen (Integer, Integer)

What's really baffling to me is I feel like this is how it *should* look--that the whole point of the state monad is to *not* have to explicitly pass the StdGen to randomR. What am I doing wrong? _______________________________________________ Beginners mailing list Beginners@haskell.org mailto:Beginners@haskell.org http://mail.haskell.org/cgi-bin/mailman/listinfo/beginners

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

Thanks! That makes sense On 2/12/16 8:26 AM, Marcin Mrotek wrote:

Hello,

You have this type error:

Couldn't match expected type ‘StateT StdGen Data.Functor.Identity.Identity Integer’ with actual type ‘g1 -> (Integer, g1)’

It's because you're trying to use a `(s -> (a,s))` function "unwrapped" where GHC expects it "wrapped" in a StateT. The type of `state` from Control.Monad.State (mtl package) is:

state :: (s -> (a, s)) -> m a

so it could solve your mismatch, turning `g1 -> (Integer, g1)` into `StateT StdGen Identity Integer`, like:

np <- state (randomR (1, maxRandPlayers))

Alternatively, if you don't want to use mtl, you can use `StateT`s constructor directly. It's type is:

StateT :: (s -> m (a, s)) -> StateT s m a

so you'd have to compose `randomR` with `return` (or `pure`) first:

np <- StateT (return . randomR (1, maxRandPlayers))

Best regards, Marcin Mrotek _______________________________________________ Beginners mailing list Beginners@haskell.org http://mail.haskell.org/cgi-bin/mailman/listinfo/beginners