- Haskell-Cafe - Haskell.org

Need Control.Monad.State
by michael rice 28 Jul '10

28 Jul '10

How do I import Control.Monad.State? I see this note in http://en.wikibooks.org/wiki/Haskell/Understanding_monads/State Note: in some package systems used for GHC, the Control.Monad.State module is in a separate package, usually indicated by MTL (Monad Transformer Library). Michael ===============rand.hs ============== import Control.Monad.State type GeneratorState = State StdGen rollDie :: GeneratorState Int rollDie = do generator <- get let( value, newGenerator ) = randomR (1,6) generator put newGenerator return value ======================= [michael@localhost ~]$ ghci rand GHCi, version 6.12.1: http://www.haskell.org/ghc/ :? for help Loading package ghc-prim ... linking ... done. Loading package integer-gmp ... linking ... done. Loading package base ... linking ... done. Loading package ffi-1.0 ... linking ... done. rand.hs:1:7: Could not find module `Control.Monad.State': Use -v to see a list of the files searched for. Failed, modules loaded: none. Prelude>

4 8

fine control of bytestring streaming
by Alberto G. Corona 27 Jul '10

27 Jul '10

Hi, I´m streaming content using lazy bytestrings in a web application. The problem is that the output comes in huge blobs (presumably of 32k) one at a time. This is good for some purposes, but not for console-like interfaces or runtime log visualization. (one of my purposes is to web-alize ghci. I know that this can be done without streaming the standard output, but I want it that way). I can not control when the IO bytestring primitives flush the data to the stream, because this is not part of my application (nor does it should do it). I suppose here that it is in a chunk by chunk basis. But maybe I misinterpreted something or everything. I use the package hack with the server hack-handler-simpleserver. this serves does Data.ByteString.hPut the bytestring content... I use mappend to add content to the stream. mappend uses "append" which uses. "foldrChunks" which seems not to compact two bytestrings in max size chunks . So I do not know what happens. The question is: are there some way to control bytestring streaming?. Can It be done without the stream handler? Alberto.

2 1

Haskell at CERN
by Alexey Khudyakov 27 Jul '10

27 Jul '10

Hello I'm wondering is there any haskellers in CERN? Given quality and usability of software there must be people looking for better alternatives.

1 0

transform function for a GADT
by Ozgur Akgun 27 Jul '10

27 Jul '10

Café, I've tried several things already, but I am not including any of them for now. My question is, how would you define the 'transform' function for a GADT, say the one in the linked gist: http://gist.github.com/492364 (also attached to this e-mail) To be concise, I want 'transform' to apply the transformation function (its first parameter) to every immidiate child of its second parameter as long as the types match. Similar to what the 'tranform' function of Uniplate does for normal ADTs. (But just one level, so I guess it is more similar to the 'descend' of Uniplate. See http://hackage.haskell.org/packages/archive/uniplate/1.2.0.1/doc/html/Data-… ) I think I got closest to a sensible solution using multi-param type classes, and defining many instances for different combinations of ExprType's but still there were problems. Waiting for suggestions and/or insights. Best, Ozgur Akgun

2 1

Actors and message-passing a la Erlang
by Yves Parès 27 Jul '10

27 Jul '10

Hello ! I've been studying Erlang and Scala, and I was wondering if someone has already implemented an actors and message passing framework for concurrent and distributed programs in Haskell.

10 12

An experimental Zipper using Thrists and first-class Labels. Help or thoughts?
by Brandon Simmons 27 Jul '10

27 Jul '10

I had the idea for a simple generic Zipper data structure that I thought would be possible to implement using type-threaded lists provided by Gabor Greif's thrist package: http://hackage.haskell.org/package/thrist ...and the fclabels package by Sebastiaan Visser, Erik Hesselink, Chris Eidhof and Sjoerd Visscher: http://hackage.haskell.org/package/fclabels It would (ideally) work as follows: - the zipper would consist simply of a tuple: (type threaded list of constructor sections , current "context") - in the type threaded list we store functions (constructor with hole -> complete constructor), so the "one hole context" is represented as a lambda expression where the free variable will be filled by the current "context" (the snd of the tuple) - we "go down" through our structure by passing to our `moveTo` function a first-class label corresponding to the constructor we want to descend into. `moveTo` uses this both as a "getter" to extract the next level down from the current level, and as a "setter" to form the lambda expression which acts as the "constructor with a piece missing" - "going up" means popping the head off the thrist and applying it to the current context, making that the new context, exiting the zipper would be a fold in the same manner After throwing together a quick attempt I realized that I'm not sure if it would be possible to make the `moveUp` function type-check and be usable. I'm still new to GADTs, existential types, template haskell etc. and am stuck. Here is the code I wrote up, which doesn't currently compile: ---------------------------------- START CODE ------------------------------- {-# LANGUAGE TypeOperators, GADTs #-} module ZipperGenerator ( viewC --lets user pattern match against context , moveTo , moveUp , genZippers , zipper , unzipper , (:->) , ZipperGenerator , Zipper ) where -- these provide the secret sauce import Data.Record.Label import Data.Thrist import Language.Haskell.TH type ZipperGenerator = [Name] -> Q [Dec] -- the Template Haskell function that does the work of generating -- first-class labels used to move about the zipper: genZippers :: ZipperGenerator genZippers = mkLabels -- hide the innards: newtype Zipper t c = Z (Thrist (->) c t, c) -- returns the current "context" (our location in the zipper) for pattern -- matching and inspection: viewC :: Zipper t c -> c viewC (Z(_,c)) = c -- takes a first-class label corresponding to the record in the current context -- that we would like to move to: moveTo :: (c :-> c') -> Zipper t c -> Zipper t c' moveTo lb (Z(thr,c)) = Z (Cons (\a-> set lb a c) thr , get lb c) -- backs up a level in the zipper, returning `Nothing` if we are already at the -- top level: moveUp :: Zipper t c -> Maybe (Zipper t b) moveUp (Z (Nil,_)) = Nothing moveUp (Z (Cons f thr,c)) = Just $ Z (thr, f c) -- create zipper with focus on topmost constructor level: zipper :: t -> Zipper t t zipper t = Z (Nil,t) -- close zipper unzipper :: Zipper t c -> t unzipper (Z(thr,c)) = undefined --foldThrist ($) id thr c ---------------------------------- END CODE ------------------------------- Thanks, Brandon Simmons http://coder.bsimmons.name/blog/

2 3

monoids and monads
by John Lato 27 Jul '10

27 Jul '10

Hello, I was wondering today, is this generally true? instance (Monad m, Monoid a) => Monoid (m a) where mempty = return mempty mappend = liftM2 mappend I know it isn't a good idea to use this instance, but assuming that the instance head does what I mean, is it valid? Or more generally is it true for applicative functors as well? I think it works for a few tricky monads, but that's not any sort of proof. I don't even know how to express what would need to be proven here. Any resources for how I could develop a means to reason about this sort of property? Thanks, John

5 7

Is 'flip' really necessary?
by aditya siram 27 Jul '10

27 Jul '10

Eta-reducing is nice, and sometimes it makes code more readable. But 'flip' is one of those functions that always seems to hinder rather than help readability, conversely factoring out flip always makes code easier to comprehend. I don't see a need for its existence - maybe I'm missing something and more experienced Haskellers can comment. -deech

15 19

1st attempt at parallelizing
by Günther Schmidt 27 Jul '10

27 Jul '10

Hi all, I'm spidering web pages, the implementation currently is synchronous. I'd like to parallelize this for speed-up, ie. get up to 6 pages in parallel and recycle those threads. Now I have come across good examples for this on the web before, but I doubt I'd find it again right away. I'd appreciate some good pointers. Günther

3 5

Re: Instances for Set of Functor, Traversable?
by oleg＠okmij.org 27 Jul '10

27 Jul '10

Lennart Augustsson wrote: > Try to make Set an instance of Functor and you'll see why it isn't. > It's very annoying. And yet the very simple, and old solution works. http://okmij.org/ftp/Haskell/types.html#restricted-datatypes We just properly generalize Functor, so that all old functors are new functors. In addition, many more functors become possible, including Set. In general, we can have functors fmap' :: (C1 a, C2 b) => (a -> b) -> f a -> f b Incidentally, even an Integer may be considered a functor: we can define the fmap' operation fitting the above signature, where the constraint C1 a is a ~ Integer. Although the use of OverlappingInstances is not required, the extension leads to the nicest code; all old functors just work. {-# LANGUAGE MultiParamTypeClasses, FlexibleInstances #-} {-# LANGUAGE OverlappingInstances #-} module FunctorEx where import Control.Monad import Data.Set as S class Functor' f a b where fmap' :: (a -> b) -> f a -> f b -- The default instance: -- All ordinary Functors are also extended functors instance Functor f => Functor' f a b where fmap' = fmap -- Now define a functor for a set instance (Ord a, Ord b) => Functor' S.Set a b where fmap' = S.map -- Define a degenerate functor, for an integer newtype I a = I Integer deriving Show instance Functor' I Integer Integer where fmap' f (I x) = I $ f x -- tests -- Lists as functors test_l = fmap' (+10) [1,2,3,4] -- [11,12,13,14] -- Sets as functors test_s = fmap' (\x -> x `mod` 3) $ S.fromList [1,2,3,4] -- fromList [0,1,2] -- Integer as functor test_i = fmap' (* (6::Integer)) $ I 7 -- I 42

3 2