- Haskell-Cafe - Haskell.org

sinkfork using conduit 3.0
by grant 25 Mar '12

25 Mar '12

I am trying to update sinkfork to use conduits 3.0. The difficulty I am having is trying to figure out how to push a value into a SinkM constructor. Any ideas? Here is the old! conduits code 2.0: https://gist.github.com/1975383 Thanks in advance. Grant

1 0

Alex character sets
by Omari Norman 24 Mar '12

24 Mar '12

I'm wrinting (or at least, trying to write) an Alex lexer. Say I want to group character ranges together and specify that I want a character that is *not* in any of those ranges. Example --- { module Main where } %wrapper "posn-bytestring" $lowAscii = \x00-\x1f -- unprintable stuff $lowSymbol = \x21-\x2f -- exclamation point to solidus $space = \x20 $special = [^$lowAscii $lowSymbol] lex :- $special { Word } { data Token = Word AlexPosn ByteString.ByteString } When I run this through Alex, I get "Prelude.Enum.Char.pred: bad argument". What am I doing wrong? How would I accomplish this sort of thing? Thanks. --Omari

1 1

ANNOUNCE: options, an easy-to-use command-line option parser
by John Millikin 24 Mar '12

24 Mar '12

Hackage: http://hackage.haskell.org/package/options Home page: https://john-millikin.com/software/haskell-options/ API reference: https://john-millikin.com/software/haskell-options/reference/haskell-option… The options package lets library and application developers easily work with command-line options. The following example is a full program that can accept two options, --message and --quiet: ---------------------------------------------------------- {-# LANGUAGE TemplateHaskell #-} import Options defineOptions "MainOptions" $ do stringOption "optMessage" "message" "Hello world!" "A message to show the user." boolOption "optQuiet" "quiet" False "Whether to be quiet." main :: IO () main = runCommand $ \opts args -> do if optQuiet opts then return () else putStrLn (optMessage opts) ---------------------------------------------------------- ---------------------------------------------------------- $ ./hello Hello world! $ ./hello --message='ciao mondo' ciao mondo $ ./hello --quiet $ ---------------------------------------------------------- In addition, this library will automatically create documentation options such as --help and --help-all: ---------------------------------------------------------- $ ./hello --help Help Options: -h, --help Show option summary. --help-all Show all help options. Application Options: --message A message to show the user. --quiet Whether to be quiet. ----------------------------------------------------------

1 0

Un-memoization
by Victor Miller 23 Mar '12

23 Mar '12

I was writing a Haskell program which builds a large labeled binary tree and then does some processing of it, which is fold-like. In the actual application that I have in mind the tree will be *huge*. If the whole tree is kept in memory it would probably take up 100's of gigabytes. Because of the pattern of processing the tree, it occurred to me that it might be better (cause much less paging) if some large subtrees could be replaced by thunks which can either recalculate the subtree as needed, or write out the subtree, get rid of the references to it (so it can be garbage collected) and then read back in (perhaps in pieces) as needed. This could be fairly cleanly expressed monadically. So does anyone know if someone has created something like this? Victor

4 3

Open-source projects for beginning Haskell students?
by Brent Yorgey 23 Mar '12

23 Mar '12

Hi everyone, I am currently teaching a half-credit introductory Haskell class for undergraduates. This is the second time I've taught it. The last time, for their final project I gave them the option of contributing to an open-source project; a couple groups took me up on it and I think it ended up being a modest success. So I'd like to do it again this time around, and am looking for particular projects I can suggest to them. Do you have an open-source project with a few well-specified tasks that a relative beginner (see below) could reasonably make a contribution towards in the space of about four weeks? I'm aware that most tasks don't fit that profile, but even complex projects usually have a few "simple-ish" tasks that haven't yet been done just because "no one has gotten around to it yet". If you have any such projects, I'd love to hear about it! Here are a few more details: * The students will be working on the projects from approximately the end of this month through the end of April. * By "relative beginner" I mean someone familiar with the material listed here: http://www.cis.upenn.edu/~cis194/lectures.html and just trying to come to terms with Applicative and Monad. They definitely do not know much if anything about optimization/profiling, GADTs, the mtl, or Haskell-programming-in-the-large. (Although part of the point of the project is to teach them a bit about programming-in-the-(medium/large)). * What I would hope from you is a willingness to exchange email and/or chat with the student(s) over the course of the project, to give them a bit of guidance/mentoring. I am certainly willing to help on that front, but of course I probably don't know much about your particular project. thanks! -Brent

10 13

Parallelism causes space leaks
by Yavuz Yetim 23 Mar '12

23 Mar '12

Hi, For the code below, where it says "HERE" in comments, if I remove the part after `using` the code works fine. However, with this version it causes a Stack space overflow (if allowed uses GBs of memory). You just need to input a file with around 1M lines each having something like "Int Value: 3 @x". What is wrong with adding the parList to this code? (Same thing happens for parMap, and parListChunk, etc) Yavuz import System.IO import System.Environment import System.IO.Error import Control.Parallel import Control.Parallel.Strategies import Control.Monad import Data.Binary as DB import Data.Binary.Put import Data.Word import Data.Maybe import qualified Data.ByteString.Lazy as B import Text.Regex.TDFA import Text.Regex.Base.Context main = do { args <- getArgs; x <- getLines (head args); mapM_ writeMaybeIntBinary ((map perLineOperator x) `using` parList rdeepseq); -- HERE return (); } writeMaybeIntBinary :: Maybe Word32 -> IO () writeMaybeIntBinary Nothing = return (); writeMaybeIntBinary (Just intB) = do { B.hPut stdout (runPut (putWord32host intB));}; getLines :: FilePath -> IO [String] getLines = liftM lines . readFile perLineOperator :: String -> Maybe Word32 perLineOperator line = let { getIntStr :: String -> String; getIntStr "" = ""; getIntStr line = let { matches = (line =~ "Int Value: (-?[0-9]*) .*" :: [[String]]); } in if matches == [] then "" else (last (head matches)); } in let { intStr = (getIntStr line); } in if intStr == "" then Nothing else Just (read intStr :: Word32)

1 0

[ANNOUNCE] vector-conduit
by Jared Hance 23 Mar '12

23 Mar '12

I've released a small (one-module) library [1] inspired by functions in Data.Conduit.List like sourceList, consumeList, and take. The most recent (by a few days) depends on conduit-0.3.*, but vector-conduit-0.2.1.0 depends on conduit-0.2.*. Notably, however, the two have different APIs (0.3 has an API more consistent with other vector operations). Use of any versions below 0.2.1 is discouraged as consumeVector is notably less efficient since in those versions it relies on a dlist. This should be useful for replacing consumeList with consumeVector, since consumeVector is much more efficient for reading in a large amount of values with, at this point, very little code overhead for using a Vector rather than a list. The library uses Data.Vector.Generic so unboxed, boxed, immutable, and mutable vectors are all supported, along with conduits for converting between mutable and immutable. [1] http://hackage.haskell.org/package/vector-conduit-0.3.0.0

2 4

haskell-platform vs macports
by Warren Harris 23 Mar '12

23 Mar '12

I assume that many haskell users out there on macs who are also users of macports, and I bet they've hit this same issue that I've hit numerous times. The problem is that there are 2 incompatible versions of libiconv -- one that ships with the mac, and one that's built with macports and that many macports-compiled libraries depend on. Work-arounds have been documented in numerous places (notably here: http://blog.omega-prime.co.uk/?p=96) but if you are trying to link with a macports-compiled libraries that depends on /opt/local/lib/libiconv.2.dylib, your only alternative seems to be building ghc from source in order to avoid the incompatibility. I hit this problem while trying to build a foreign-function interface to libarchive. So I built ghc from scratch -- in fact I built the whole haskell-platform. This was relatively painless, and fixed the libiconv problem. However, it brings me to the real point of my message, which is that the version of haskell-platform available via macports is way out of date (2009.2.0.2 http://www.macports.org/ports.php?by=name&substr=haskell-platform) I'm wondering whether the haskell-platform team has considered maintaining a macports version of the haskell-platform for us mac users in addition to the binary install that's available now. This would avoid the incompatibilities such as this nagging one with libiconv. Perhaps it's just a matter of maintaining template versions of the port files? Warren Undefined symbols for architecture x86_64: "_locale_charset", referenced from: _localeEncoding in libHSbase-4.3.1.0.a(PrelIOUtils.o) "_iconv_close", referenced from: _hs_iconv_close in libHSbase-4.3.1.0.a(iconv.o) (maybe you meant: _hs_iconv_close) "_iconv", referenced from: _hs_iconv in libHSbase-4.3.1.0.a(iconv.o) (maybe you meant: _hs_iconv, _hs_iconv_open , _hs_iconv_close ) "_iconv_open", referenced from: _hs_iconv_open in libHSbase-4.3.1.0.a(iconv.o) (maybe you meant: _hs_iconv_open) ld: symbol(s) not found for architecture x86_64 collect2: ld returned 1 exit status cabal: Error: some packages failed to install: Libarchive-0.1 failed during the building phase. The exception was: ExitFailure 1

2 2

Puzzled about inlining and specialise inline under ghc -O2
by Rafal Kolanski 22 Mar '12

22 Mar '12

Dear Haskell-Cafe, I'm computing a histogram of a bunch of symbols with up to 8 bits of information each, stored in a unboxed vector of Word8. The histogram is represented as an unboxed vector of Int with size 2^bits. I compute the histogram by folding an increment function. The problem: depending on what types and what annotations I give to the increment and histogram function (see below), the GC gets through a different amount of memory. I'm using GHC 7.0.3 and -O2. I'd like to better understand how and why the optimisation does/doesn't kick in. Here are the functions with the most generic types I can think of: import qualified Data.Vector.Unboxed as UV import qualified Data.Vector.Unboxed.Mutable as UMV import Control.Monad.Primitive (PrimMonad, PrimState) increment :: (PrimMonad m, UMV.Unbox a, Num a, Integral b) => UMV.MVector (PrimState m) a -> b -> m (UMV.MVector (PrimState m) a) increment v x = do n <- UMV.read v (fromIntegral x) UMV.write v (fromIntegral x) (n+1) return v histogram :: (Integral a, UMV.Unbox a) => Int -> UV.Vector a -> UV.Vector Int histogram bitsPerSym v = runST $ do a <- UMV.replicate (2^bitsPerSym) (0::Int) a' <- UV.foldM' increment a v UV.unsafeFreeze a' Running my test load, I get: total alloc = 33,206,568 bytes Looking at the core, ghc is not specialising the functions, even if I tell it to inline them. So let's brutally change the types to be as specific as I need for my application: increment :: UMV.MVector s Int -> Word8 -> ST s (UMV.MVector s Int) histogram :: Int -> UV.Vector Word8 -> UV.Vector Int result: total alloc = 19,581,152 bytes and if I put INLINE pragmas for both functions: 16,952,512 bytes I should be able to achieve the same effect with SPECIALISE INLINE pragmas, right? Let's try that: {-# SPECIALISE INLINE increment :: UMV.MVector s Int -> Word8 -> ST s (UMV.MVector s Int) #-} {-# SPECIALISE INLINE histogram :: Int -> UV.Vector Word8 -> UV.Vector Int #-} result: 33,139,856 bytes (GHC can't figure out application of the first rule, giving: Warning: RULE left-hand side too complicated to desugar) So unfortunately my most generic form won't work here, I need to specialise increment to be in ST (which sucks, because I want it to work for both IO and ST): increment :: (UMV.Unbox a, Num a, Integral b) => UMV.MVector s a -> b -> ST s (UMV.MVector s a) {-# SPECIALISE INLINE increment :: UMV.MVector s Int -> Word8 -> ST s (UMV.MVector s Int) #-} result: 17,016,192 bytes This is very close to the most specific function instantiations and INLINE, but: - I've lost being generic between ST and IO - it's still a little bigger than the specific instances + INLINE So my questions are: what is going on? Can I have genericity between ST and IO while keeping the low GC usage? How come SPECIALISE INLINE does not give the same result as specific instances + INLINE? Obviously, for this example, I don't really *need* increment to work inside IO, since I'm using runST... but I want to understand what is going on. Profiling Haskell performance and memory usage has always been difficult for me. Much thanks in advance, Rafal Kolanski.

1 0

Munich Haskell Meeting
by Heinrich Hördegen 22 Mar '12

22 Mar '12

Dear all, once again, it's time for our monthly Haskell get-together in Munich. On Wed, 28 Feb 2012, at 19h30, we will meet at Cafe Puck (near the universities). Everybody is invited. If you plan to join, please go to: http://www.haskell-munich.de/dates and add yourself. This will help to reserve tables. Have a nice time until Wednesday, Heinrich -- -- hoerdegen(a)funktional.info www.funktional.info --

1 0