March 2008 - Haskell-Cafe

Announce: Decimal arithmetic package
by Paul Johnson 30 Mar '08

30 Mar '08

I've just uploaded version 0.1.0 of a decimal arithmetic package to Hackage. Decimal numbers are stored as an Integral mantissa and a Word8 exponent, where the number stored is mantissa * 10^(-exponent). In other words the exponent is the number of decimal places stored. There are also routines for partitioning a value such that the sum of the parts is equal to the original value. This is useful in financial arithmetic. This came out of my ongoing work on an AMQP binding for Haskell. Version 0-10 of the AMQP spec includes decimal fractions defined in this way for financial applications, so I thought I'd make a proper job of it. A darcs repository will be set up shortly. Paul.

1 0

announce: Glome.hs raytracer
by Jim Snow 30 Mar '08

30 Mar '08

I have recently posted a haskell port of my ocaml raytracer, Glome: http://syn.cs.pdx.edu/~jsnow/glome/ It supports spheres and triangles as base primitives, and is able to parse files in the NFF format used by the standard procedural database (http://tog.acm.org/resources/SPD/) It uses a bounding interval heirarchy acceleration structure, so it can render fairly complicated scenes in a reasonable amount of time. Shadows and reflections are supported, but not specular highlights or refraction. It's still slower than the ocaml version, but at least they're in the same ballpark (and a good part of that difference may be inefficiencies in my BIH traversal). I would welcome any advice on making it go faster or use less memory. I compile the program with "ghc Glome.hs --make -fasm -O2 -threaded -fglasgow-exts -funbox-strict-fields -fbang-patterns -fexcess-precision -optc-ffast-math -optc-O2 -optc-mfpmath=sse -optc-msse2". (I assume the -optc options don't do anything unless you compile via C.) Here are some of my current concerns: -Multi-core parallelism is working, but not as well as I'd expect: I get about a 25% reduction in runtime on two cores rather than 50%. I split the default screen size of 512x512 into 16 blocks, and run "parMap" on those blocks with a function that turns the screen coordinates of that block into a list of (x,y,r,g,b) tuples that get drawn as pixels to the screen through OpenGL by the original thread. -Memory consumption is atrocious: 146 megs to render a scene that's a 33k ascii file. Where does it all go? A heap profile reports the max heap size at a rather more reasonable 500k or so. (My architecture is 64 bit ubuntu on a dual-core amd.) -Collecting rendering stats is not easy without global variables. It occurs to me that it would be neat if there were some sort of write-only global variables that can be incremented by pure code but can only be read from within monadic code; that would be sufficient to ensure that the pure code wasn't affected by the values. The sorts of things I'm looking for are the number of calls to "trace" per image, the number of BIH branches traversed and ray/triangle and ray/sphere intersections per pixel. (Disclaimer: I don't really fully understand monads, so I may be oblivious to an obvious solution.) -Is there a fast way to cast between Float and Double? I'm using Float currently, and the only reason is because that's what the OpenGL api expects. I'd like to be able to use either representation, but the only way to cast that I've found so far is "float_conv x = fromRational(toRational x)", which is too slow. thanks, -jim

10 15

announce: hgdbmi (GDB/MI interface)
by Evan Martin 30 Mar '08

30 Mar '08

Just in case someone else needed this, here you go: GDB/MI lets programs drive GDB. It can be used, for example, by GDB frontends. This module wraps attaching GDB to a process and parsing the (surprisingly complicated) GDB/MI output. http://hackage.haskell.org/cgi-bin/hackage-scripts/package/hgdbmi

1 0

ANNOUNCE: xmonad 0.7 released
by Don Stewart 29 Mar '08

29 Mar '08

http://xmonad.org The xmonad dev team is pleased to announce xmonad 0.7! The headlines: The 0.7 release of xmonad provides several improvements over 0.6, including improved integration with GNOME, more flexible "rules", various stability fixes, and of course, many new and interesting features in the extension library (general support for window decorations, utf8 support, scratch pad terminals, pointer control) and more! New GNOME support: Active, automated support for GNOME utilities. We know our users often like to use GNOME menus, tools and status bars, and we'd like to announce that xmonad actively supports GNOME! Extensions for communicating with and utilising gnome utilities come in the library suite, as well as extensive documentation and support. For more information see the GNOME/xmonad integration page on the wiki. A period of active development: In the past year, the xmonad development team received contributions from over 60 developers, making xmonad one of the largest window manager development teams around, and dwarfing other tiling window manager projects. Yet, at the same time, the core code base remains at around 1000 lines of code, with another 7000 lines in the extension library -- a significant achievment! Change logs: http://haskell.org/haskellwiki/Xmonad/Notable_changes_since_0.6 http://xmonad.org/changelog-0.7.txt http://xmonad.org/changelog-xmc-0.7.txt About: xmonad is a tiling window manager for X. Windows are arranged automatically to tile the screen without gaps or overlap, maximising screen use. Window manager features are accessible from the keyboard: a mouse is optional. xmonad is extensible in Haskell, allowing for powerful customisation. Custom layout algorithms, key bindings and other extensions may be written by the user in config files. Layouts are applied dynamically, and different layouts may be used on each workspace. Xinerama is fully supported, allowing windows to be tiled on several physical screens. Features: * Very stable, fast, small and simple. * Automatic window tiling and management * First class keyboard support: a mouse is unnecessary * Full support for tiling windows on multi-head displays * Full support for floating, tabbing and decorated windows * Full support for Gnome and KDE utilities * XRandR support to rotate, add or remove monitors * Per-workspace layout algorithms * Per-screens custom status bars * Compositing support * Powerful, stable customisation and reconfiguration * Large extension library * Excellent, extensive documentation * Large, active development team, support and community Get it! Information, screenshots, documentation, tutorials and community resources are available from the xmonad home page: http://xmonad.org The 0.7 release, and its dependencies, are available from hackage.haskell.org: http://hackage.haskell.org/cgi-bin/hackage-scripts/package/xmonad xmonad packages are available in the package systems of at least: Debian, Gentoo, Arch, Ubuntu, OpenBSD, NetBSD, FreeBSD, Gobo, NixOS, Source Mage, Slackware and 0.7 packages will appear in coming days (some are already available). On the fly updating to xmonad 0.7 is supported. You should be able to upgrade to xmonad 0.7 from 0.6 and earlier, transparently, without losing your session. Load the new code with mod-q and enjoy. Extensions: xmonad comes with a huge library of extensions (now more than 7 times the size of xmonad itself), contributed by viewers like you. Extensions enable pretty much arbitrary window manager behaviour to be implemented by users, in Haskell, in the config files. For more information on using and writing extensions see the webpage. The library of extensions is available from hackage: http://hackage.haskell.org/cgi-bin/hackage-scripts/package/xmonad-contrib Full documentation for using and writing your own extensions: http://xmonad.org/contrib.html This release brought to you by the xmonad dev team: Spencer Janssen Don Stewart Jason Creighton David Roundy Brent Yorgey Devin Mullins Braden Shepherdson Roman Cheplyaka Lucas Mai Featuring code contributions from over 60 developers: Aaron Denney Adam Vogt Alec Berryman Alex Tarkovsky Alexandre Buisse Andrea Rossato Austin Seipp Bas van Dijk Ben Voui Brandon Allbery Chris Mears Christian Thiemann Clemens Fruhwirth Daniel Neri Daniel Wagner Dave Harrison David Glasser David Lazar Dmitry Kurochkin Dominik Bruhn Dougal Stanton Eric Mertens Ferenc Wagner Gwern Branwen Hans Philipp Annen Ivan Tarasov Jamie Webb Jeremy Apthorp Joachim Breitner Joachim Fasting Joe Thornber Joel Suovaniemi Juraj Hercek Justin Bogner Kai Grossjohann Karsten Schoelzel Klaus Weidner Mathias Stearn Mats Jansborg Matsuyama Tomohiro Michael Fellinger Michael Sloan Miikka Koskinen Neil Mitchell Nelson Elhage Nick Burlett Nicolas Pouillard Nils Anders Danielsson Peter De Wachter Robert Marlow Sam Hughes Shachaf Ben-Kiki Shae Erisson Simon Peyton Jones Stefan O'Rear Tom Rauchenwald Valery V. Vorotyntsev Will Farrington Yaakov Nemoy timthelion As well as the support of many others on the #xmonad and #haskell IRC channels, and the wider Haskell and window manager communities. Thanks to everyone for their support!

1 0

Haddock Help Required
by Dominic Steinitz 29 Mar '08

29 Mar '08

What should I be using for the file name for the read-interface option in haddock? Trying to use the file on www.haskell.org gives this: dom@lagrange:~/asn15/asn1> haddock -html -o hdoc Pretty.hs -B /usr/lib/ghc-6.8.2 --optghc="-fglasgow-exts" --read-interface=http://www.haskell.org/ghc/docs/latest/html/libraries/base,http://www.haskell.org/ghc/docs/latest/html/libraries/base/base.haddock haddock: internal Haddock or GHC error: http://www.haskell.org/ghc/docs/latest/html/libraries/base/base.haddock: openBinaryFile: does not exist (No such file or directory) But if I download the file and try and use it locally, I get this dom@lagrange:~/asn15/asn1> haddock -v -html -o hdoc Pretty.hs -B /usr/lib/ghc-6.8.2 --optghc="-fglasgow-exts" --read-interface=http://www.haskell.org/ghc/docs/latest/html/libraries/base,base.haddock Warning: Cannot read base.haddock: "Magic number mismatch: couldn't load interface file: base.haddock" Skipping this interface. Warning: main:Language.ASN1: could not find link destinations for: GHC.Enum.Enum GHC.Base.Eq GHC.Base.Ord GHC.Show.Show GHC.Num.Integer Language.ASN1.ComponentIndex Language.ASN1.Tagged GHC.Base.String Data.Maybe.Maybe Language.ASN1.Octet Data.Map.Map Warning: main:Pretty: could not find link destinations for: Text.PrettyPrint.HughesPJ.Doc Pretty.Pretty Pretty.PrettyVal Warning: main:ConstrainedType: could not find link destinations for: Data.Maybe.Maybe GHC.Base.Bool GHC.Num.Integer Pretty.Pretty Pretty.PrettyVal GHC.Base.String GHC.Base.Eq GHC.Show.Show GHC.Base.Int GHC.Word.Word8 GHC.Base.Ord Data.Monoid.Monoid Data.Binary.Strict.BitPut.BitPut GHC.Real.Integral Control.Monad.State.Class.MonadState Data.ByteString.Internal.ByteString Control.Monad.Error.Class.MonadError GHC.Base.Char Data.Binary.Strict.BitGet.BitGet Any help would be appreciated. Dominic.

2 2

Monad instance for Data.Set, again
by Henning Thielemann 29 Mar '08

29 Mar '08

The blog article http://www.randomhacks.net/articles/2007/03/15/data-set-monad-haskell-macros describes a variant of the Monad class which allows to restrict the type of the monadic result, in order to be able to make Data.Set an instance of Monad (requiring Ord constraint for the monadic result). The same problem arises for container data structures with restricted element types, where the element type restriction depends on the implementation of the container structure (such as UArray). It would be cumbersome to have several class parts, even more, type constraints in type signatures may reveal implementation details. E.g. constraint (Container c x y z) might be needed for a 'zipWith' function, whereas (Container c y x z) is needed if you use 'zipWith' with swapped arguments. Here is another approach that looks tempting, but unfortunately does not work, and I wonder whether this can be made working. module RestrictedMonad where import Data.Set(Set) import qualified Data.Set as Set class AssociatedMonad m a where class RestrictedMonad m where return :: AssociatedMonad m a => a -> m a (>>=) :: (AssociatedMonad m a, AssociatedMonad m b) => m a -> (a -> m b) -> m b instance (Ord a) => AssociatedMonad Set a where instance RestrictedMonad Set where return = Set.singleton x >>= f = Set.unions (map f (Set.toList x)) GHC says: RestrictedMonad.hs:21:13: Could not deduce (Ord b) from the context (RestrictedMonad Set, AssociatedMonad Set a, AssociatedMonad Set b) arising from use of `Data.Set.unions' at RestrictedMonad.hs:21:13-22 Probable fix: add (Ord b) to the class or instance method `RestrictedMonad.>>=' In the definition of `>>=': >>= x f = Data.Set.unions (map f (Data.Set.toList x)) In the definition for method `RestrictedMonad.>>=' In the instance declaration for `RestrictedMonad Set'

5 5

Glome raytracer bug: bad output with -O2 -fasm
by Jim Snow 29 Mar '08

29 Mar '08

I was trying to get Blinn highlights working with my raytracer, and kept getting ugly artifacts. After trying a bunch of things, I finally compiled without -O2, and the artifacts went away. Here's what I mean: http://syn.cs.pdx.edu/~jsnow/glome/Glome.hs-noartifact.png http://syn.cs.pdx.edu/~jsnow/glome/Glome.hs-artifact.png Here's the offending code, run "./make" and "./run" and you should see the artifacts if your setup is the same as mine. (Requires OpenGL.) http://syn.cs.pdx.edu/~jsnow/glome/glome.hs-0.2-bug.tar.gz The artifacts also go away if I use -fvia-C. It doesn't seem to matter whether I use Floats or Doubles in the rendering code. The artifacts also show up with -O1. Have I stumbled across a known compiler bug? Or perhaps an OpenGL bug? (The bug could, of course, be in my code, but then one might expect to get the same erroneous output every time regardless of compiler flags.) To reiterate, I'm using ghc 8.6.2. thanks, -jim

2 2

Sound typeable via type families?
by Dan Doel 28 Mar '08

28 Mar '08

Hello, Recently, someone (I forgot who, exactly) mentioned in #haskell this article: http://okmij.org/ftp/Haskell/types.html#unsound-typeable where Oleg demonstrates that Typeable makes the type system unsound, and one can get back unsafeCoerce (which is used in the implementation, I suspect). However, this evening, Ryan Ingram showed how to construct a safe Typeable using GADTs. The problem with that, of course, is that GADTs aren't open, so one can only have a predefined number of Typeables. However, it occurred to me that data families don't have that problem, so I took a shot and came up with the following: {-# LANGUAGE TypeFamilies, ScopedTypeVariables, GADTs , MultiParamTypeClasses, FlexibleInstances, OverlappingInstances #-} class Typeable t where data TypeRep t :: * typeRep :: TypeRep t instance Typeable Int where data TypeRep Int = TInt typeRep = TInt instance Typeable Integer where data TypeRep Integer = TInteger typeRep = TInteger instance Typeable () where data TypeRep () = TUnit typeRep = TUnit data TEq a b where Refl :: TEq a a class (Typeable a, Typeable b) => TEquality a b where teq :: TypeRep a -> TypeRep b -> Maybe (TEq a b) instance (Typeable a) => TEquality a a where teq _ _ = Just Refl instance (Typeable a, Typeable b) => TEquality a b where teq _ _ = Nothing cast :: forall a b. (TEquality a b) => a -> Maybe b cast a = do Refl <- teq ta tb return a where ta :: TypeRep a ta = typeRep tb :: TypeRep b tb = typeRep Which, somewhat to my surprise, actually works. And, of course, if one tries to pull off Oleg's trick: newtype Oleg a = Oleg { unOleg :: a } instance Typeable (Oleg a) where TypeRep (Oleg a) = ... typeRep = ?? Well, it won't work, because TypeRep () ~/~ TypeRep (Oleg a). However, obviously, this depends on overlapping instances (if there's some other way, I'd be happy to know; if type inequality contexts are available, I wasn't able to find them), and I've heard that type families don't play well with overlap. Does that not apply to data families? Will this construction still work in 6.10 and beyond? Also, this doesn't seem to be a suitable basis for Dynamic. Trying to extend the GADT solution presented resulted in errors unless incoherent instances were turned on (clearly not a good sign), and even then, it didn't actually work. Is it possible to do better, and come away with something that will actually work for Dynamic, and be sound? Is there some other trick waiting to pull unsafeCoerce out of this setup (seems doubtful, as it isn't used in the code, and the type families folks have no doubt done plenty of work to ensure soundness)? Comments appreciated. -- Dan

3 3

What Does This Message Mean? (Re: [Haskell-cafe] compilation succeeds -- execution fails)
by Jason Dusek 28 Mar '08

28 Mar '08

The message unknown symbol `___stginit_cedictzm0zi1zi1_DataziCharziCEDICTziMatter_' says that it can't find the initializer for` Data.Char.CEDICT.Matter` in `cedict-0.1.1` (this is 'z-encoding', if I remember correctly). So, the odd thing is, that is not the part with the C FFI stuff in it -- that's in `Data.Char.CEDICT.Lists`. Is there a usual thing that is wrong when the initializer can not be found? -- _jsn

2 1

[gsoc] mingw64 ghc port
by Bulat Ziganshin 28 Mar '08

28 Mar '08

Hello haskell-cafe, it's probably a bit too late, but i recalled that there is one project that will be very useful - it's porting ghc to mingw64 platform, allowing it to generate 64-bit windows platforms. may be someone will find it interesting -- Best regards, Bulat mailto:Bulat.Ziganshin@gmail.com

1 0