December 2008 - Haskell-Cafe

Re: [Haskell] ANN: HLint 1.0
by Neil Mitchell 22 Dec '08

22 Dec '08

Hi Lennart, > It would be nice if HLint didn't suggest things that it will object to > in the next round. > Like > > LLVM/Core/CodeGen.hs:176:1: Eta reduce > Found: > applyArgs f g = apArgs 0 f g > Why not: > applyArgs f = apArgs 0 f The idea is to specify things one step at a time, so the user learns from the hints, as well as fixing them. However, in the case of eta reduction, doing several reductions at a time does make sense. I'll add it to the todo list. Thanks Neil

1 0

Removing/Uninstalling cabal packages ?
by Laurent Giroud 22 Dec '08

22 Dec '08

Hi, I have been doing a few experiments with cabal packages lately and I wish to uninstall these to return to a cleaner package base. However, there doesn't seem to be a "cabal uninstall" command and reading the documentation at http://www.haskell.org/ghc/docs/latest/html/Cabal/index.html does not tell much about how to do that manually. Note that what I wish to do is both delete the installed files and the cabal database so that the uninstalled packages does not show anymore when running "cabal list --installed". Would anyone know how to do that ? Thanks in advance. Regards, Laurent -- Laurent Giroud ml(a)niaow.com

3 2

RE: [Haskell-cafe] Trouble with FFI on Windows2000...
by Serge LE HUITOUZE 22 Dec '08

22 Dec '08

Hi Kyra (I tried to email you directly, but, for some reason, my mail bounced...), Hi Haskellers, > > dlltool -d cproj1.def -l libcproj1.a > > ghc --make testFFI_2.hs -optl-lcproj1 -optl-L. > > [...] > > This leads to an error box with message: > > "DLL (null) not found on specified path" > > dlltool -d cproj1.def -l libcproj1.a -D cproj1.dll Thanks very much for this answer, short and effective (an ideal one)! Just an additional question, just for curiosity: Does anyone know why this "-D" option doesn't appear in the following two pages? - http://www.haskell.org/haskellwiki/GHC/Using_the_FFI#Setting_up_your_build_… - http://www.haskell.org/haskellwiki/GHC:FAQ#GHC_on_Windows (parag. 1.4.3.4) Is this because the behavior I got is Win2K specific (hence very few people run into in 2008)? Thanks again Kyra for your precious help. --Serge

1 0

Yampa vs. Reactive
by Tony Hannan 22 Dec '08

22 Dec '08

Hello, Can someone describe the advantages and disadvantages of the Yampa library versus the Reactive library for functional reactive programming, or point me to a link. Thanks, Tony P.S. It is hard to google for Yampa and Reactive together because "reactive" as in "function reactive programming" always appears with Yampa

5 17

ANN: llvm-0.4.0.1
by Lennart Augustsson 22 Dec '08

22 Dec '08

We have now released version 0.4.0.1 of the Haskell LLVM bindings. (This release that is quite incompatible with the old 0.0.2 release.) LLVM is a virtual machine and the bindings allow you to generate code for this virtual machine. This code can then be executed by a JIT or written to a file for further processing by the LLVM tools. The LLVM bindings has two layers. You can either use the low level bindings that is just the same as the C bindings for the LLVM. This level is quite unsafe as there is very little type checking. The recommended way is a high level binding (somewhat less complete) which eliminates many errors by leveraging the Haskell type system. A simple example, generating code for a function that adds two numbers and then calling it from Haskell: import Data.Int import LLVM.Core import LLVM.ExecutionEngine cgplus :: CodeGenModule (Function (Int32 -> Int32 -> IO Int32)) cgplus = createFunction InternalLinkage $ \ x y -> do r <- add x y ret r main = do ioplus <- simpleFunction cgplus let plus = unsafePurify ioplus print $ plus 20 22 Enjoy! Bryan O'Sullivan Lennart Augustsson

1 0

Records and associated types
by Taru Karttunen 21 Dec '08

21 Dec '08

Hello What is the correct way to transform code that uses record selection with TypeEq (like HList) to associated types? I keep running into problems with overlapping type families which is not allowed unless they match. The fundep code: class Select rec label val | rec label -> val instance TypeEq label label True => Select (Label label val :+: rest) label val instance (Select tail field val) => Select (any :+: tail) field val And a conversion attempt: class SelectT rec label where type S rec label instance TypeEq label label True => SelectT (Label label val :+: rest) label where type S (Label label val :+: rest) label = val instance (SelectT tail field) => SelectT (any :+: tail) field where type S (any :+: tail) field = S tail field which fails with: Conflicting family instance declarations: type instance S (Label label val :+: rest) label -- Defined at t.hs:19:9 type instance S (any :+: tail) field -- Defined at t.hs:23:9 How is it possible to get the TypeEq constraint into the type family? Attached is a complete example that illustrates the problem. - Taru Karttunen

3 2

ANN: Data.List.Split
by Brent Yorgey 21 Dec '08

21 Dec '08

I am pleased to announce the release of Data.List.Split, which provides a wide range of strategies and a unified combinator framework for splitting lists with respect to some sort of delimiter. It strives to be flexible yet simple. If you've ever wished there was a simple 'split' function you could grab from Data.List, this is the package for you---no matter which of the seventeen* slightly different ways to split a list you actually need! Just use one of the prepackaged splitting strategies, or build your own with the combinator library. Get it from Hackage: cabal install split http://hackage.haskell.org/cgi-bin/hackage-scripts/package/split Get the source: darcs get http://code.haskell.org/~byorgey/code/split Get excited: http://byorgey.wordpress.com/2008/12/21/datalistsplit/ -Brent * Actual number may vary.

1 0

Logos of Other Languages
by Ashley Yakeley 21 Dec '08

21 Dec '08

I browsed around a bit for logos from other languages... Python http://www.python.org/images/python-logo.gif The "snake pair" is visually interesting while still remaining simple. The typeface is unusual and yet clean and humanistic. The logo is only slightly marred by the "TM". Overall, elegant and appealing. A. Caml http://caml.inria.fr/styles/modern/title-en.gif This is actually more of a heading than a logo: it continues to the right edge of the page to close the oval. The typeface is readable, but the text is a bit verbose, spelling out "The Caml Language" with a visual nod to its ML roots. B+. Ruby http://www.ruby-lang.org/images/logo.gif An ordinary book typeface with an ordinary picture of an ordinary ruby. And the strapline "A Programmer's Best Friend" is vague and uninspired. Bland, but at least inoffensive. C. Perl http://upload.wikimedia.org/wikipedia/en/e/e0/Programming-republic-of-perl.… I'm not sure if this is "the Perl logo" or "O'Reilly's Perl logo", but it's ugly, busy, difficult to read and a bit obscure. The title "PROGRAMMING REPUBLIC OF" also lends an air of snotty pretension. D. Apple Dylan http://osteele.com/projects/images/Dylan.logo-thumb.png Arty, abstract, vaguely Hermitian, but a bit corporate. Easy to make a website favicon from, but the actual word "Dylan" is not strictly included. A-. All of these get one thing right that the current and most of the proposed Haskell logos do not: they don't make any reference to the syntax of the language itself. Doing so seems to miss the point of a logo: it's supposed to appeal visually, rather than semantically. So I'd like to see some submissions that don't use lambdas. -- Ashley Yakeley Seattle, WA

16 19

Re: [Haskell] ANN: Hoogle with more libraries
by Neil Mitchell 21 Dec '08

21 Dec '08

Hi Bob, > 1. Searching using a package name that isn't all lower case results in > nothing (e.g. (a -> b) -> f a -> f b +InfixApplicative gives no results, > while (a -> b) -> f a -> f b +infixapplicative gives 2). Yes, if you do +InfixApplicative it assumes you mean only in the module InfixApplicative, not the package. Module is upper case, package starts with lowercase. > 2. Searching with the +package syntax seems to remove all other possible > results. Perhaps it would be nice to show results from the default library > set below. Add +default to search the default too. > If you get the package name wrong (i.e. specify a package that hoogle can't > see), it would be nice for it to report something like google does -- maybe > you meant xyz. Definately! The whole area needs an overhall. Thanks Neil

1 0

Windows vs. Linux x64
by Bartosz Wójcik 21 Dec '08

21 Dec '08

Hi Everybody, while working on my resent project I've noticed that my code seems to be faster under Windows than under Linux x64. More exactly this was an AI game evaluator that ran on given parameters. There was no IO performed. I've run 3 lots of test on both systems and stored some figures. It was physicaly the same PC. 1st lot WinXP total time = 27.18 secs (1359 ticks @ 20 ms) total alloc = 5,788,242,604 bytes (excludes profiling overheads) Linux total time = 34.44 secs (1722 ticks @ 20 ms) total alloc = 11,897,757,176 bytes (excludes profiling overheads) 2nd lot WinXP total time = 63.96 secs (3198 ticks @ 20 ms) total alloc = 13,205,507,148 bytes (excludes profiling overheads) Linux total time = 80.76 secs (4038 ticks @ 20 ms) total alloc = 27,258,694,888 bytes (excludes profiling overheads) 3rd lot WinXP total time = 207.10 secs (10355 ticks @ 20 ms) total alloc = 44,982,716,780 bytes (excludes profiling overheads) Linux total time = 267.58 secs (13379 ticks @ 20 ms) total alloc = 92,307,482,416 bytes (excludes profiling overheads) I've used the same compile and runtime options for both. I've tried to run with -H option, but this didn't improve anything. Is this common behaviour? Does anybody know what can be the reason? regards, Bartek

7 11