- Haskell-Cafe - Haskell.org

Eliminating phantom type
by Timotej Tomandl 09 Apr '17

09 Apr '17

Hi, I have stumbled upon the Control-Monad-ST2 package by Kevin Backhouse, I have decided to expand the code with a run function and replace IO with ST s; and as of now I have removed the functions working on arrays and conversion to IO for the sake of simplicity. Here is the code in its modified form with newtype wrappers removed: {-# LANGUAGE RankNTypes #-} {-# LANGUAGE DeriveFunctor #-} -- Copyright 2013 Kevin Backhouse. -- Copyright 2017 Timotej Tomandl. Modifications import Data.STRef import Control.Monad.ST import Control.Applicative import Control.Monad newtype ST2 r w s a = ST2 { unwrapST2 :: ST s a } deriving (Functor) instance Monad (ST2 r w s) where return x = ST2 $ return x (ST2 m) >>= f = ST2 $ do x <- m unwrapST2 (f x) instance Applicative (ST2 r w s) where pure = return (<*>) = ap -- | This function checks that the sub-computation is polymorphic in -- both type parameters. This means that the sub-computation does not -- read or write any state from the enclosing context. {-# INLINE pureST2 #-} pureST2 :: (forall r w. ST2 r w s a) -> ST2 r' w' s a pureST2 m = m -- | This function checks that the computation is polymorphic in -- both parameters and then returns a pure value {-# INLINE runST2 #-} runST2 :: (forall r w s. ST2 r w s a) -> a runST2 m=runST $ unwrapST2 m -- | This function checks that the sub-computation is polymorphic in -- the @w@ type parameter. This means that the sub-computation does -- not write any state from the enclosing context (but read-only -- operations are permitted). {-# inline readOnlyST2 #-} readOnlyST2 :: (forall w. ST2 r w s a) -> ST2 r w' s a readOnlyST2 m = m -- | Mutable reference. 'ST2Ref' is actually just a newtype of an -- 'STRef', but the @r@ and @w@ type parameters allow the read and -- write dependencies to be tracked by the type system. newtype ST2Ref r w s a = ST2Ref (STRef s a) -- | Create a new reference. The @r@ and @w@ type parameters of the -- reference are unified with the 'ST2' monad to indicate that new -- state is created in the enclosing context. {-# INLINE newST2Ref #-} newST2Ref :: a -> ST2 r w s (ST2Ref r w s a) newST2Ref x = ST2 $ do r <- newSTRef x >>= \ var -> return var return (ST2Ref r) -- | Read a reference. The @w@ type parameter of the reference is not -- unified with the 'ST2' monad to indicate that this access is -- read-only. {-# INLINE readST2Ref #-} readST2Ref :: ST2Ref r w s a -> ST2 r w' s a readST2Ref (ST2Ref r) = ST2 $ readSTRef r -- | Write to a reference. The @w@ type parameter of the reference is -- unified with the 'ST2' monad to indicate that state is written in -- the enclosing context. {-# INLINE writeST2Ref #-} writeST2Ref :: ST2Ref r w s a -> a -> ST2 r w s () writeST2Ref (ST2Ref r) x = ST2 $ writeSTRef r x -- | Modify a reference. {-# INLINE modifyST2Ref #-} modifyST2Ref :: ST2Ref r w s a -> (a -> a) -> ST2 r w s () modifyST2Ref (ST2Ref r) f = ST2 $ modifySTRef r f But as you can see, now all my types got tainted by s even though sharing the same r or w implies sharing the same s. Is there a way how to hide s from the type signatures, but still preserve an ability to write runST2 without resorting to IO/RealWorld? I have tried writing such a function and failed. Timotej Tomandl

3 3

State monad exit
by p75213＠gmail.com 08 Apr '17

08 Apr '17

Hi, I am playing around with the State monad and queues. At the moment I have the following code: {-# LANGUAGE ViewPatterns, FlexibleContexts #-} module Main where import Criterion.Main import Control.Monad.State.Lazy import Data.Maybe (fromJust) import Data.Sequence ((<|), ViewR ((:>))) import qualified Data.Sequence as S -------------------------------------------------------- data Queue a = Queue { enqueue :: [a], dequeue :: [a] } deriving (Eq, Show) -- adds an item push :: a -> Queue a -> Queue a push a q = Queue (a:enqueue q) (dequeue q) pop :: Queue a -> Maybe (a, Queue a) pop q = if null (dequeue q) then go $ Queue [] (reverse (enqueue q)) else go q where go (Queue _ []) = Nothing go (Queue en (x:de)) = Just (x, Queue en de) queueTst :: Int -> Queue Int -> Queue Int queueTst 0 q = q queueTst n q | even n = queueTst (n - 1) (push (100 + n) q) | otherwise = queueTst (n - 1) (if popped == Nothing then q else snd (fromJust popped)) where popped = pop q ------------------------------------------------------------- pushS :: a -> S.Seq a -> S.Seq a pushS a s = a <| s pushS' :: a -> State (S.Seq a) (Maybe a) pushS' a = do s <- get put (a <| s) return Nothing pushS'' :: a -> State (S.Seq a) (Maybe a) pushS'' a = get >>= (\g -> put (a <| g)) >> return Nothing popS :: S.Seq a -> Maybe (a, S.Seq a) popS (S.viewr -> S.EmptyR) = Nothing popS (S.viewr -> s:>r) = Just (r,s) popS' :: State (S.Seq a) (Maybe a) popS' = do se <- get let sl = popS'' se put $ snd sl return $ fst sl where popS'' (S.viewr -> S.EmptyR) = (Nothing, S.empty) popS'' (S.viewr -> beg:>r) = (Just r, beg) queueTstS :: Int -> S.Seq Int -> S.Seq Int queueTstS 0 s = s queueTstS n s | even n = queueTstS (n - 1) (pushS (100 + n) s) | otherwise = queueTstS (n - 1) (if popped == Nothing then s else snd (fromJust popped)) where popped = popS s queueTstST :: Int -> State (S.Seq Int) (Maybe Int) queueTstST n = if (n > 0) then if even n then pushS' (100 + n) >> queueTstST (n - 1) else popS' >> queueTstST (n - 1) else return Nothing main1 :: IO () main1 = defaultMain [ bench "Twin Queue" $ whnf (queueTst 550) (Queue [500,499..1] []) , bench "Sequence Queue" $ whnf (queueTstS 550) (S.fromList [500,499..1]) , bench "State Queue" $ whnf (runState (queueTstST 550)) (S.fromList [500,499..1]) ] -------------------------------------------------------------------------------------------------------------------------------------------------- In the function "queueTstST" is there a way to exit while retaining the last "Maybe value" rather than with "Nothing"?

1 0

LambdaConf standard registration opens
by Salas Sanchez-Bennasar 08 Apr '17

08 Apr '17

http://lambdaconf.us/events/2017/lcusc.html LambdaConf is one of the largest functional programming conferences in the world, focused on foundational concepts (such as lambda calculus, type theory, category theory), functional programming languages (Haskell, Scala, PureScript, F#, Clojure, Elixir, and more), new languages and approaches in functional programming, functional programming libraries, and related areas such as dependent types, proof systems, parallel array programming, functional GPU processing, and more. Standard registration is now open at Eventbrite: https://www.eventbrite.com/e/lambdaconf-2017-registration-29839799644 A complete list of sessions is online, including the following highlights: * 6 keynotes, including one on using Haskell to model life on the blockchain, one on how programmers brains work, and one on the axes of abstraction; * 8 leap workshops, including an in-depth workshop on category theory, one on optics, and one on building front-ends using PureScript Halogen; * 12 hop workshops, including one introducing Coq, another diving into recursion schemes, and another teaching functional data processing; * 9 de novo sessions, including on one formally specified digital logic, one on algebraic and monadic composability for distributed computing, and one on type-level REST programming; * 21 educational sessions, including one on Free monads, one on high-performance Haskell, one on codata and corecursion, one on FP on Android using Frege, and another on dependently-typed programming in Haskell; * 21 inspire sessions, including one on dependent-pairs, one on generative design, and one on type singletons. This is more than 110 hours of content, with enough variety and difficulty level for everyone. Speakers are book authors, contributors to open source libraries, researchers, and practicing functional programmers. The conference takes place in Boulder, Colorado, from May 25 - 27, at the University of Colorado Boulder in the foothills of the Rocky Mountains. Conference registration includes all sessions (including workshops), locally-catered breakfast and lunch on Thursday, Friday, and Saturday; a conference dinner on Friday night; a choice from a variety of networking activities on Sunday; free childcare and STEM workshops for children; all-gender restrooms and handicapped-accessible venues. Videos from some previous events can be found here: https://www.youtube.com/ channel/UCEtohQeDqMSebi2yvLMUItg; and here: https://www.youtube.com/ watch?v=JxC1ExlLjgw&list=PLE7tQUdRKcybh21_zOg8_y4f2oMKDHpUS The conference is preceded by commercial training opportunities on Monday - Tuesday, including: * Up & Running with Elixir & Phoenix, by Brooklyn Zelenka * Applied Haskell, by Michael Snoyman * Advanced FP in Scala, by John A. De Goes * Introduction to FP, by David Koontz * Mastering Apache Spark, by Pawel Szulc * Mastering Elm, by Isaac Shapira On Wednesday, several free mini-conferences are co-located with LambdaConf: * PureScript Conf 2017 * Introduction to FP with Haskell For more information on LambdaConf, the commercial training workshops, or the mini-conferences, please visit the website: http://lambdaconf.us/events/2017/lcusc.html

1 0

Downsides of the Prompt Monad
by Michael Litchard 08 Apr '17

08 Apr '17

Could someone outline for me the downsides of using the Prompt monad?

5 7

[ANN] etc-0.0.0 Declarative configuration spec for Haskell projects
by Román González 08 Apr '17

08 Apr '17

Is my pleasure to announce a new library for managing configuration values on Haskell projects. Some features: * Have a versioned spec of all values your application can accept * Provides an API for gathering values from multiple sources (files, overwrite files, cli arguments, environment variables) and then composing them into a single configuration map * Gives a sane precedence over sources of the configuration value sources * Provides inspection utilities to understand why the configuration is the way it is * Provides an API that abstracts away the source of configuration values and allows easy casting into record types your application or other libraries understand * Supports both configuration files in JSON or YAML (when cabal flags are used) * Dynamically generates CLI inputs from spec configuration file Any feedback, or comments feel free to create a ticket or send me a tweet over at @romanandreg Cheers. Roman Gonzalez.-

2 2

Re: [Haskell-cafe] What is a safe Haskell data type to store and manipulate Money values
by Doug McIlroy 07 Apr '17

07 Apr '17

> REFERENCES TO LITERATURE PLEASE Another domain with pertinent literature is proportional representation in legislative bodies. Sorry, I have no particular references to offer. Doug

1 0

Package maintainership takeover: quickcheck-instances
by Oleg Grenrus 07 Apr '17

07 Apr '17

The quickcheck-instances provides QuickCheck type-classes' instances for Haskell Platform packages. Unfortunately, author and current maintainer Antoine Latter is quite unresponsive. There is handful of PR open [2], and I had to use Hackage Trustee powers to allow new versions of dependencies (to keep e.g. Stackage nightly going on). I have tried to contact him to offer a co-maintainership help (at 14.12.2016). Cheers, Oleg Grenrus - [1]: http://hackage.haskell.org/package/quickcheck-instances - [2]: https://github.com/aslatter/qc-instances/pulls

1 0

What is a safe Haskell data type to store and manipulate Money values?
by Sandeep Cr 07 Apr '17

07 Apr '17

Hello all, What is a good data type for storing and manipulating monetary values in a typical web application with payment processing? What are the pros and cons of various options like Data.Scientific, Rational and Decimal?

13 24

Re: [Haskell-cafe] [Haskell] Haskell Tools
by amindfv＠gmail.com 07 Apr '17

07 Apr '17

Moving this conversation to haskell-cafe! > El 6 abr 2017, a las 04:15, Boldizsár Németh <nboldi(a)caesar.elte.hu> escribió: > > Thanks for asking, I just added a note that you need to install the haskell-tools-daemon package to use the Atom plugin for Haskell Tools. > > If you want to run refactorings from a script or just without an editor, you can use haskell-tools-cli as well. > >> On 2017.04.06. 10:40, Ivan Lazar Miljenovic wrote: >> Is there support for anything except Atom at this time? >> >> I'm also unsure of what I need to install on my local machine if I >> want to try it out (do I need the daemon? or is the cli tool >> sufficient?). >> >>> On 6 April 2017 at 17:03, Boldizsár Németh <nboldi(a)caesar.elte.hu> wrote: >>> Dear Haskellers, >>> >>> I'm happy to announce a new tool for refactoring Haskell source code. >>> Check it out: http://haskelltools.org/ >>> >>> Best Regards, >>> Boldizsár Németh >>> _______________________________________________ >>> Haskell mailing list >>> Haskell(a)haskell.org >>> http://mail.haskell.org/cgi-bin/mailman/listinfo/haskell > > _______________________________________________ > Haskell mailing list > Haskell(a)haskell.org > http://mail.haskell.org/cgi-bin/mailman/listinfo/haskell

2 1

Re: [Haskell-cafe] [ANN] etc-0.0.0 Declarative configuration spec for Haskell projects
by Geraldus 07 Apr '17

07 Apr '17

Román González: > Was too excited delivering the newspaper that I forgot to add it to the > email: > > Github repo: https://github.com/roman/Haskell-etc >

1 0