September 2009 - Haskell-Cafe

Trying to reduce memory costs of String duplicates
by Günther Schmidt 06 Sep '09

06 Sep '09

Hi all, I'm reading in a data of 216k records into a map of Key, Values Pairs, the values being strings. As it happens out of 216k String values there really are only about 6.6k distinct string values, so I could save a lot of RAM if I was able to "insert" only actually *new* string values into the map and use references to (string) values that already are in memory instead. Is there a container that would, if I wanted to insert an element, return a pair of either the previously inserted, equal value and the container unchanged, or the new, previously unknown value and the new container amended by that element? Günther

5 6

Haskell Weekly News: Issue 129 - September 5, 2009
by Brent Yorgey 05 Sep '09

05 Sep '09

--------------------------------------------------------------------------- Haskell Weekly News http://sequence.complete.org/hwn/20090905 Issue 129 - September 05, 2009 --------------------------------------------------------------------------- Welcome to issue 129 of HWN, a newsletter covering developments in the [1]Haskell community. The [2]Haskell Symposium was a great success, with many [3]interesting talks and a good discussion on the future of Haskell. Watch this space for links to video from the Symposium as it becomes available! Announcements HStringTemplate 0.6.2. Sterling Clover [4]announced some new features in the [5]HStringTemplate library, including simple quasiquotation; proper Unicode support; creation of groups from hierarchies of directories; separators applied within iterated template application; depthwise chained iterated template application; generalized encoding functions; and more. fclabels-0.4.0 - First class accessor labels. Sebastiaan Visser [6]announced a new release of the [7]fclabels package, straight from ICFP in Edinburgh. The package provides first-class labels which act as fully composable, bidirectional record fields, as well as support for automatically generating them from record types. vty-4.0.0.1 released. Corey O'Connor [8]announced release 4.0.0.1 of [9]vty, a terminal UI library. This release brings a number of important fixes, features, and performance enhancements, including a completely rewritten output backend; efficient, "scanline rasterization" style output span generator; terminfo based display terminal implementation; improved Unicode support; 256 color support; and more. haskell-src-exts-1.1.4. Niklas Broberg [10]announced the release of [11]haskell-src-exts-1.1.4, a package for Haskell source code manipulation. The experimental code in Language.Haskell.Annotated{.*} has changed quite a lot, although the stable portion of the package interface has not changed. Significantly, the package now includes an exact-printer which allows round-tripping between parsing and pretty-printing to be the identity. Next BostonHaskell meeting: September 16th at MIT (32G-882). Ravi Nanavati [12]announced the September meeting of the Boston Area Haskell Users' Group, to be held Wednesday, September 16th from 7pm - 9pm. As usual, it will be held in the MIT CSAIL Reading Room (32-G882, on the 8th floor of the Gates Tower of the MIT's Stata Center at 32 Vassar St in Cambridge, MA). The featured speaker will be Edward Kmett, who will be presenting the second part of his monoids and parsing presentation: "A Parallel Parsing Trifecta: Iteratees, Parsec, and Monoids". lenses -- Simple Functional Lenses. Job Vranish [13]announced the release of [14]lenses, a simple but powerful implementation of function lenses (aka functional references/accessors). This library provides a convenient way to access and update the elements of a structure. It is very similar to Data.Accessors, but simpler, a bit more generic and has fewer dependencies. Dutch HUG: meeting next week (September 11th) in Utrecht. Tom Lokhorst [15]invited functional programmers in The Netherlands to the [16]Dutch Haskell User Group, meeting Friday, September 11 at 19:00 in the [17]Booth Hall of the Utrecht University Library. Thomas (noknok) will be talking about his system for doing propositional logic in Haskell. Pedro will give an introductory talk about generic programming, and Sean will talk about xformat, a library for extensible and type-safe formatting with scanf- and printf-like functions. There is also still space for short 5-minute lighting talk about something related to Haskell or functional programming; contact Tom if you're interested. moe html combinator. Jinjing Wang [18]announced the release of [19]moe, a DSL for generating HTML. jail-0.0.1 - Jailed IO monad. Sebastiaan Visser [20]announced the first release of the [21]jail package, a jailed IO monad that can restrict filesystem access for your code. scion 0.1. Thomas Schilling [22]announced the first release of [23]Scion, a Haskell library that aims to implement those parts of a Haskell IDE which are independent of a particular front-end. Scion is based on the GHC API and Cabal. It provides both a Haskell API and a server for non-Haskell clients such as Emacs and Vim. Blog noise [24]Haskell news from the [25]blogosphere. Blog posts from people new to the Haskell community are marked with >>>, be sure to welcome them! * Don Stewart (dons): [26]DEFUN 2009: Multicore Programming in Haskell Now!. * Bryan O'Sullivan: [27]Slides from my CUFP 2009 keynote talk. * LHC Team: [28]Yet another unfair benchmark.. * Alex McLean: [29]Hackpact documentation. * >>> Jeff Foster: [30]Exploring Haskell's List Functions. * Don Stewart (dons): [31]Parallel Programming in Haskell: A Reading List. * David Amos: [32]Finite fields, part 2. * >>> Jeff Foster: [33]Debugging in Haskell. * apfelmus: [34]Fun with Morse Code. * Ketil Malde: [35]Parsing ints. * Alex McLean: [36]Hackpact. * Greg Bacon: [37]Finding duplicates with Perl and Haskell. * Colin Adams: [38]Selecting software for a replacement for this website. * Don Stewart (dons): [39]Haskell Popularity Rankings: September 2009. * Mikael Vejdemo-Johansson: [40][Stanford] MATH 198: Category Theory and Functional Programming. * Jeff Heard: [41]HDR imaging library for Haskell based on pfsutils. * David Amos: [42]Extension fields. * Magnus Therning: [43]Trying to work out iteratees. Quotes of the Week * benmachine: ho hum. I understand both your positions. but i don't understand mine, now :( * ksf: agda is actually a secret mindwar-weapon of the illuminati, who want to wrack your nerves with excessively big symbol sets requiring a keyboard with 10 modifier keys. just like APL. * Axman6: does anyone else think that C++ looks like a dead fish? (C++<) * Cale: The difference between Many Worlds and Copenhagen is a garbage collector ;) * apfelmus: Lambda Fu, form 72 - three way dragon zip: 'averages3 xs = zipWith3 avg xs (drop 1 xs) (drop 2 xs); where avg a b c = (a+b+c) / 3' About the Haskell Weekly News New editions are posted to [44]the Haskell mailing list as well as to [45]the Haskell Sequence and [46]Planet Haskell. [47]RSS is also available, and headlines appear on [48]haskell.org. To help create new editions of this newsletter, please see the information on [49]how to contribute. Send stories to byorgey at cis dot upenn dot edu. The darcs repository is available at darcs get [50]http://code.haskell.org/~byorgey/code/hwn/ . References 1. http://haskell.org/ 2. http://www.haskell.org/haskell-symposium/2009/ 3. http://www.haskell.org/haskell-symposium/2009/schedule.html 4. http://article.gmane.org/gmane.comp.lang.haskell.cafe/63108 5. http://hackage.haskell.org/package/HStringTemplate 6. http://article.gmane.org/gmane.comp.lang.haskell.cafe/63082 7. http://hackage.haskell.org/package/fclabels 8. http://article.gmane.org/gmane.comp.lang.haskell.cafe/63072 9. http://hackage.haskell.org/package/vty 10. http://article.gmane.org/gmane.comp.lang.haskell.cafe/63068 11. http://hackage.haskell.org/package/haskell-src-exts 12. http://article.gmane.org/gmane.comp.lang.haskell.cafe/63067 13. http://article.gmane.org/gmane.comp.lang.haskell.cafe/63059 14. http://hackage.haskell.org/package/lenses 15. http://article.gmane.org/gmane.comp.lang.haskell.cafe/63000 16. http://www.haskell.org/haskellwiki/Dutch_HUG 17. http://www.uu.nl/EN/library/contact/university_library/zaalverhuur/Pages/de… 18. http://article.gmane.org/gmane.comp.lang.haskell.cafe/62865 19. http://hackage.haskell.org/package/moe 20. http://article.gmane.org/gmane.comp.lang.haskell.cafe/62839 21. http://hackage.haskell.org/package/jail 22. http://article.gmane.org/gmane.comp.lang.haskell.cafe/62820 23. http://hackage.haskell.org/package/scion 24. http://planet.haskell.org/ 25. http://haskell.org/haskellwiki/Blog_articles 26. http://donsbot.wordpress.com/2009/09/05/defun-2009-multicore-programming-in… 27. http://www.serpentine.com/blog/2009/09/04/slides-from-my-cufp-2009-keynote-… 28. http://lhc-compiler.blogspot.com/2009/09/yet-another-unfair-benchmark.html 29. http://yaxu.org/hackpact-documentation/ 30. http://www.fatvat.co.uk/2009/09/exploring-haskells-list-functions.html 31. http://donsbot.wordpress.com/2009/09/03/parallel-programming-in-haskell-a-r… 32. http://haskellformaths.blogspot.com/2009/09/finite-fields-part-2.html 33. http://www.fatvat.co.uk/2009/09/debugging-in-haskell.html 34. http://apfelmus.nfshost.com/fun-with-morse-code.html 35. http://blog.malde.org/index.php/2009/08/31/parsing-ints/ 36. http://yaxu.org/hackpact/ 37. http://feedproxy.google.com/~r/gbacon/~3/aaZePz6Gv2Q/finding-duplicates-wit… 38. http://colina.demon.co.uk/?q=node/753 39. http://donsbot.wordpress.com/2009/08/29/haskell-popularity-rankings-septemb… 40. http://blog.mikael.johanssons.org/archive/2009/08/stanford-math-198-categor… 41. http://vis.renci.org/jeff/2009/08/28/hdr-imaging-library-for-haskell-based-… 42. http://haskellformaths.blogspot.com/2009/08/extension-fields.html 43. http://therning.org/magnus/archives/735 44. http://www.haskell.org/mailman/listinfo/haskell 45. http://sequence.complete.org/ 46. http://planet.haskell.org/ 47. http://sequence.complete.org/node/feed 48. http://haskell.org/ 49. http://haskell.org/haskellwiki/HWN 50. http://code.haskell.org/~byorgey/code/hwn/

1 0

Re: Problem on existential type.
by John Lato 05 Sep '09

05 Sep '09

Also, the two definitions for liftGW are exactly equivalent. You have: liftGW (GridWidget label) f = f label liftGW (GridWidget textView) f = f textView The only difference between the two is name to which the data parameter to GridWidget is bound, which doesn't change the meaning at all. You should properly have liftGW (GridWidget widget) f = f widget There's no way to determine if that widget is a label or textView, which is the whole point of existential types. Cheers, John > From: Miguel Mitrofanov <miguelimo38(a)yandex.ru> > Subject: Re: [Haskell-cafe] Problem on existential type. > To: Magicloud Magiclouds <magicloud.magiclouds(a)gmail.com> > > Your data type GridWidget doesn't have a parameter, yet you use it > like it has one. > >> data GridWidget = forall widget. (WidgetClass widget) => GridWidget >> widget > ^ > | > NB:-------------+ > >> liftGW :: (GridWidget widget) -> (widget -> t) -> t >

1 0

Re: [Haskell-cafe] haskelldb + sqlite problem.
by Colin Paul Adams 05 Sep '09

05 Sep '09

>>>>> "Magicloud" == Magicloud Magiclouds <magicloud.magiclouds(a)gmail.com> writes: Magicloud> Hi, I am using haskelldb and Magicloud> haskelldb-hdbc-sqlite3. Well, I finally got the source Magicloud> compiled and ran, I got this error: App: user error Magicloud> (SQL error: SqlError {seState = "", seNativeError = 21, Magicloud> seErrorMsg = "prepare 74: SELECT subject,\n Magicloud> timestamp\nFROM notes as T1\nORDER BY timestamp DESC: Magicloud> library routine called out of sequence"}) Any clue what Magicloud> I should check? Did you get this working? If so, what was the problem and how did you go about resolving it. I have the identical problem. I had the database code working fine, but then I added a state monad into the monad stack for the program, and now I get this problem. I see that John Goerzen suggested it might be a result of reading the data lazily. So I tried changing my import statements from import Control.Monad.State to import Control.Monad.State.Strict in case the StateT was indirectly causing the problem, but that doesn't make any difference. -- Colin Adams Preston Lancashire

1 1

How to preload the module of my own
by zaxis 05 Sep '09

05 Sep '09

I want to preload the module automatically when starting ghci. The module located in ~/work directory contains some functions i use everyday. Now i use an alias: alias ghci='ghci -i ~/money/Money.hs' which works fine. However i feel there maybe are more elegant way. thanks! -- View this message in context: http://www.nabble.com/How-to-preload-the-module-of-my-own-tp25304154p253041… Sent from the Haskell - Haskell-Cafe mailing list archive at Nabble.com.

3 3

Native windowing for glut applications on OS X.
by Lyndon Maydwell 05 Sep '09

05 Sep '09

Hi haskell-cafe. I recently updated glut through cabal to version GLUT-2.2.1.0, and where once I had native windowing, now I can only seem to use X11. Does anyone know how to use native windowing?

2 1

forkM fails
by Alberto G. Corona 04 Sep '09

04 Sep '09

Hi I need to execute a procedure not in the IO monad, but in an any monad: I defined: forkM :: Monad m=> m a -> IO ThreadId forkM proc=forkIO $ proc `seq` return() I assumed that seq will force the evaluation of proc and after, it will discard his type (m a) and return () in the IO monad.as forkIO expect. however proc is not executed Prelude> Control.Concurrent.forkIO $ print "hola" ThreadId 331 "hola" Prelude> Prelude> let forkM p=Control.Concurrent.forkIO $ p `seq` return () Prelude> forkM $ print "hola" ThreadId 493 Prelude> Any idea?. Thanks in advance

3 2

ANN: HStringTemplate 0.6.2
by Sterling Clover 04 Sep '09

04 Sep '09

I haven't sent out a release announcement for HStringTemplate in some time. In that time, there's been a host of new users, and consequently a host of feature requests and bug reports. Among the relatively recent changes: * Simple quasiquotation. * Proper unicode support (files are read in UTF-8, Data.Text templates are supported). * Creation of groups from hierarchies of directories. * Separators are applied within iterated template application. * Chained iterated template application now occurs depthwise rather than breadthwise. * Encoding functions generalized from String -> String to (Stringable a) => a -> a * Trailing newlines stripped from templates. * Initial newlines stripped after conditionals. * Functions to check templates for proper parsing, and existence of all necessary parameters and invoked templates. Current version is, as always, on Hackage: http://hackage.haskell.org/package/HStringTemplate User-maintained documentation on the wiki: http://www.haskell.org/haskellwiki/HStringTemplate Patches, bug reports, requests, and especially additional work on the documentation always welcome. Cheers, Sterl.

1 0

[Long] 'Fun' with types
by Maciej Piechotka 04 Sep '09

04 Sep '09

I'm somehow experimenting with GADT. I'm implementing FRP system and I get such error: Occurs check: cannot construct the infinite type: a = (a, b) In the pattern: CircuitSplit f g In the definition of `createChannel': createChannel (CircuitSplit f g) = let (df, cf) = createChannel f (dg, cg) = createChannel g .... in (df <<< dg, Map.map merge $ Map.unionWith (\ (First x) (Second y) -> Both x y) (Map.map First cf) (Map.map Second cf)) On code: import Control.Applicative import Control.Arrow import Control.Category (Category) import qualified Control.Category import Control.Monad import Control.Monad.Fix import Data.IORef import Data.Monoid hiding (First) import Data.Map (Map) import qualified Data.Map as Map import Data.Unique import Foreign import Foreign.C import Unsafe.Coerce -- | "CSignal" is a continous signal - i.e. signal which have a value at each -- point of time data CSignal a = CSignal (IO a) instance Functor CSignal where f `fmap` (CSignal v) = CSignal $ f <$> v instance Applicative CSignal where pure x = CSignal $ pure x (CSignal f) <*> (CSignal v) = CSignal $ f <*> v -- | "DSignal" is a discret signal - i.e. signal which is point-like. data DSignal a = DSignal (Map Unique ((a -> IO ()) -> IO ())) instance Functor DSignal where f `fmap` (DSignal l) = DSignal $ Map.map (. (. f)) l instance Monoid (DSignal a) where mempty = DSignal Map.empty (DSignal l) `mappend` (DSignal s) = DSignal $ Map.union l s -- | Convers the discret signal into a continous signal readAndCacheSignal :: a -- ^ Inital value -> DSignal a -- ^ Discret signal -> IO (CSignal a) -- ^ Continous signal readAndCacheSignal v (DSignal l) = do vr <- newIORef v Map.fold (>>) (return ()) $ Map.map ($ writeIORef vr) l return $ CSignal $ readIORef vr data Circuit a b where CircuitCSignal :: CSignal a -> Circuit () a CircuitDSignal :: DSignal a -> Circuit () (Maybe a) CircuitArr :: Kleisli IO a b -> Circuit a b CircuitJoint :: Circuit b c -> Circuit a b -> Circuit a c CircuitSplit :: Circuit a c -> Circuit b d -> Circuit (a, b) (c, d) CircuitChoice :: Circuit a c -> Circuit b d -> Circuit (Either a b) (Either c d) CircuitLoop :: Circuit (a, c) (b, c) -> Circuit a b instance Category Circuit where id = CircuitArr returnA f . g = CircuitJoint f g instance Arrow Circuit where arr f = CircuitArr $ arr f first = flip CircuitSplit returnA second = CircuitSplit returnA (***) = CircuitSplit instance ArrowChoice Circuit where left = flip CircuitChoice returnA right = CircuitChoice returnA (+++) = CircuitChoice instance ArrowLoop Circuit where loop = CircuitLoop data CircuitChannel a b where CircuitChannel :: Kleisli IO (c, a) b -> ((c -> IO ()) -> IO ()) -> CircuitChannel a b data ChannelJoin a b = First a | Second b | Both a b createChannel :: Circuit a b -> (Kleisli IO a b, Map Unique (CircuitChannel a b)) createChannel (CircuitCSignal (CSignal v)) = (Kleisli (const v), Map.empty) createChannel (CircuitDSignal (DSignal l)) = (arr $ const Nothing, Map.map (CircuitChannel (arr $ Just . snd)) l) createChannel (CircuitArr a) = (a, Map.empty) createChannel (CircuitJoint f g) = let (df, cf) = createChannel f (dg, cg) = createChannel g merge (First (CircuitChannel a r)) = CircuitChannel (dg <<< a) r merge (Second (CircuitChannel a r)) = CircuitChannel (a <<< second df) r merge (Both (CircuitChannel a r) (CircuitChannel b _)) = CircuitChannel (proc (d, v) -> do v' <- b -< (d, v) a -< (d, v')) r in (df <<< dg, Map.map merge $ Map.unionWith (\(First x) (Second y) -> Both x y) (Map.map First cf) (Map.map Second cf)) createChannel (CircuitSplit f g) = let (df, cf) = createChannel f (dg, cg) = createChannel g merge (First (CircuitChannel a r)) = CircuitChannel (proc (e, (c, d)) -> do c' <- a -< (e, c) d' <- dg -< d returnA -< (c', d')) r merge (Second (CircuitChannel b r)) = CircuitChannel (proc (e, (c, d)) -> do c' <- df -< c d' <- b -< (e, d) returnA -< (c', d')) r merge (Both (CircuitChannel a r) (CircuitChannel b _)) = CircuitChannel (proc (e, (c, d)) -> do c' <- a -< (e, c) d' <- b -< (e, d) returnA -< (c', d')) r in (df <<< dg, Map.map merge $ Map.unionWith (\(First x) (Second y) -> Both x y) (Map.map First cf) (Map.map Second cf)) createChannel (CircuitChoice f g) = let (df, cf) = createChannel f (dg, cg) = createChannel g merge (First (CircuitChannel a r)) = CircuitChannel (proc (e, c) -> case c of Left x -> a -< (e, x) Right x -> dg -< x) r merge (Second (CircuitChannel a r)) = CircuitChannel (proc (e, c) -> case c of Left x -> df -< x Right x -> a -< (e, x)) r merge (Both (CircuitChannel a r) (CircuitChannel b _)) = CircuitChannel (proc (e, c) -> case c of Left x -> a -< (e, x) Right x -> b -< (e, x)) r in (df ||| dg, Map.map merge $ Map.unionWith (\(First x) (Second y) -> Both x y) (Map.map First cf) (Map.map Second cf)) createChannel (CircuitLoop f) = let (df, cf) = createChannel f in (loop df, Map.map (\(CircuitChannel a r) -> CircuitChannel (proc (e, a) -> do rec (b, c) <- a -< (e, (a, c)) returnA -< b) r) cf) I know that it will require some sort of type voodoo but where's error now? Regards

3 3

Is this Parsec code idiomatic?
by Serge LE HUITOUZE 04 Sep '09

04 Sep '09

Hi Haskellers, I'm asking some advice on a small piece of code representing a simplified version of a treatment I need to perform. I have a line-oriented string/file, from which I want to extract only a substring of those lines starting with char '+' (the detail of the extraction is irrelevant here, I'll just return what follows the '+'). [I also simplified the "eol" parser for shorter code.] I came out with the code below. The line parser returns a "Maybe String". The complete parser return a "[Maybe String]" by mere concatenation. The main function filters the 'Nothing' with 'catMaybes'. > import Text.ParserCombinators.Parsec > import Data.Maybe > > maybePlusFile :: GenParser Char st [Maybe String] > maybePlusFile = endBy maybePlusLine eol > > maybePlusLine :: GenParser Char st (Maybe String) > maybePlusLine = try (do char('+') > result <- many (noneOf "\n") > return $ Just result) > <|> do many (noneOf "\n") > return $ Nothing > > eol = char '\n' > > selectPlus :: String -> Either ParseError [String] > selectPlus input = > case parse maybePlusFile "(input)" input of > Left e -> Left e > Right mblist -> Right $ catMaybes mblist This works as expected (or so it seems), as the ghci dump shows: > GHCi, version 6.10.1: http://www.haskell.org/ghc/ :? for help > ... > Prelude> :l selectPlus.hs > [1 of 1] Compiling Main ( selectPlus.hs, interpreted ) > Ok, modules loaded: Main. > *Main> selectPlus "abc\n+123\ndef\n+456\n" > Loading package parsec-2.1.0.1 ... linking ... done. > Right ["123","456"] > *Main> I'd like to know if this code is good style, and how you would possibly improve it. Thanks in advance. --Serge

6 5