March 2009 - Haskell-Cafe

Re: [Haskell] Google Summer of Code 2009
by Johan Tibell 10 Mar '09

10 Mar '09

On Tue, Feb 10, 2009 at 3:26 PM, Malcolm Wallace <Malcolm.Wallace(a)cs.york.ac.uk> wrote: > Gentle Haskellers, > > The Google Summer of Code will be running again this year. Once again, > haskell.org has the opportunity to bid to become a mentoring > organisation. (Although, as always, there is no guarantee of > acceptance.) > > If you have ideas for student projects that you think would benefit the > Haskell community, now is the time to start discussing them on mailing > lists of your choice. We especially encourage students to communicate > with the wider community: if you keep your ideas private, you have a > much worse chance of acceptance than if you develop ideas in > collaboration with those who will be your "customers", end-users, or > fellow-developers. This is the open-source world! > > The timeline is that Haskell.org will apply for GSoC membership between > 9-13 March, and if we are successful, students can submit applications > between 23 March - 3 April. Google is now accepting applications: http://google-opensource.blogspot.com/2009/03/google-summer-of-code-applica…

2 2

Mystified by Cabal
by Colin Paul Adams 10 Mar '09

10 Mar '09

I have just attempted Cabal-izing my program (splitting it into a library and main program as well), and I'm mystified by some problems I am having. First, when I try to build the library I get: [colin@susannah game-tree]$ runhaskell Setup build Preprocessing library game-tree-1.0.0.0... Building game-tree-1.0.0.0... Data/Tree/Game/Negascout.hs:31:0: Unrecognised pragma [1 of 2] Compiling Data.Tree.Game.Tree ( Data/Tree/Game/Tree.hs, dist/build/Data/Tree/Game/Tree.o ) Data/Tree/Game/Tree.hs:1:0: Failed to load interface for `Prelude': it is a member of package base-3.0.3.0, which is hidden That mystifies me. Googling, it appears to be a common error in the past, but none of the reasons apparently apply to my case. Incidentally, Tree.hs imports nothing. It just looks like: -- | Nodes in game trees -- Copyright 2009 Colin Adams -- -- This file is part of game-tree. -- -- Game-tree is free software: you can redistribute it and/or modify -- it under the terms of the GNU General Public License as published by -- the Free Software Foundation, either version 3 of the License, or -- (at your option) any later version. -- Game-tree is distributed in the hope that it will be useful, -- but WITHOUT ANY WARRANTY; without even the implied warranty of -- MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the -- GNU General Public License for more details. -- You should have received a copy of the GNU General Public License -- along with game-tree. If not, see <http://www.gnu.org/licenses/>. module Data.Tree.Game.Tree where -- | Nodes in a game search tree class Tree a where -- | Is this a game-terminating node (e.g. checkmate)? is_terminal :: a -> Bool -- | Heuristic value of node node_value :: a -> Int -- | Child nodes in the game tree (positions more deeply searched) children :: a -> [a] So I couldn't build my program using runhaskell, as I couldn't install the library. Instead, i tried ghc --make instead with the -idir option. It looks in the right place, but still doesn't recognise the module: Chasing modules from: *Chu-shogi.hs Generator.hs:27:7: Could not find module `Data.Tree.Game.Tree': locations searched: Data/Tree/Game/Tree.hs Data/Tree/Game/Tree.lhs dirs=../game-tree/Data/Tree/Game/Tree.hs dirs=../game-tree/Data/Tree/Game/Tree.lhs What am I doing wrong? Perhaps Cabal should be named Caballa instead. -- Colin Adams Preston Lancashire

7 12

ANN: Future 1.1.0 concurrency library
by ChrisK 10 Mar '09

10 Mar '09

Hello, As a side effect of the discussion of the new C++ future/promise features at http://lambda-the-ultimate.org/node/3221 I have implemented a Haskell package called "future" at http://hackage.haskell.org/cgi-bin/hackage-scripts/package/future This ought to do what C++ standard futures/promises do, plus a bit more. The main operation is > forkPromise :: IO a -> IO (Promise a) This sets the "IO a" operation running in a fresh thread. The eventual result can be accessed in many ways (non-blocking, blocking, blocking with timeout). > let one :: Int; one = 1 > p <- forkPromise (return (one+one)) > x <- get > y <- wait x is an Int with value 2. y is an (Either SomeException Int) with value (Right 2). The useful thing about futures, as opposed to various IVar packages, is handling the case where the forked operation ends with an exception. The exception becomes the return value of the promise. The "get" operation rethrows it, the "wait" operation returns it as (Left ...). There is also an "abort" command to kill a promise. The dead promise may then have an exceptions as its value. The "plus a bit more" than C++ is the nonblocking "addTodo" feature. This takes a continuation function from the "Either SomeException a" to an IO operation. These continuation functions get queued and they are run immediately after the the forked operation completes. Once completed any new "addTodo" continuations run immediately. These continuations allow you to race a list of action and take the first one done, or to collect the answers as they complete into a Chan. Both of those options are demonstrated in Future.hs as racePromises and forkPromises. It should be safe to use "unsafePerformIO . get" or "unsafePeformIO . wait" to get lazy access to the result, which is itself immutable once set. Cheers, Chris

3 4

Re: [Haskell-cafe] where does ghc specify ar options
by Duncan Coutts 10 Mar '09

10 Mar '09

On Mon, 2009-03-09 at 21:08 -0400, Alexy Khrabrov wrote: > Duncan -- > > On Mar 9, 2009, at 8:14 PM, Duncan Coutts wrote: > > BTW, how did you get the package installed in that location? Did it > > involve copying into a temp dir and copying again? I believe that on > > OSX, copying a .a file breaks the ar index. This is because for > > reasons > > best known to themselves Apple decided that the index is only valid if > > its timestamp is the same as the last modification time of the file. > > Oh -- and that's because instead of the supplied bootstrap.sh I used > one off some website, and it used /tmp which somehow put stuff in > there and registered it there. I then unregistered it and removed it > and redid the cabal-install, and it worked with my patched ar. Ahhh. So is there anything in Cabal or the bootstrap.sh that needs to be fixed for this problem? Duncan

1 0

where does ghc specify ar options
by Alexy Khrabrov 09 Mar '09

09 Mar '09

When bootstrapping cabal-install 0.6.2 on Mac OSX Leopard (Intel), I get a problem when linking: Linking dist/build/cabal/cabal ... ld: in /private/var/folders/mw/mwJSf7ErEa4w8nWyKyyqD++++TY/-Tmp-/zlib-0.5.0.0/dist/build/libHSzlib-0.5.0.0.a, archive has no table of contents This had happened with other ar'chives on my Mac before where it hadn't for some others. In all cases the fix was to make sure ranlib is run on the archive in question, or s is passed to ar during the creation. I have the latest XCode 312m gcc build 5490, and ghc 6.10.1 from MacPorts. In order to ensure ar cr becomes ar crs, I need to find out where in the ghc build process is it known how to create .a files? Where can an "ar cr" be changed to "ar crs" on a system-wide ghc build basis? Cheers, Alexy

2 2

Re: ANN: Future 1.1.0 concurrency library
by Neal Alexander 09 Mar '09

09 Mar '09

ChrisK wrote: > Hello, > > As a side effect of the discussion of the new C++ future/promise > features at http://lambda-the-ultimate.org/node/3221 I have implemented > a Haskell package called "future" at > > http://hackage.haskell.org/cgi-bin/hackage-scripts/package/future > > This ought to do what C++ standard futures/promises do, plus a bit > more. The main operation is > > > forkPromise :: IO a -> IO (Promise a) > > This sets the "IO a" operation running in a fresh thread. The eventual > result can be accessed in many ways (non-blocking, blocking, blocking > with timeout). > > > let one :: Int; one = 1 > > p <- forkPromise (return (one+one)) > > x <- get > > y <- wait > > x is an Int with value 2. > y is an (Either SomeException Int) with value (Right 2). > > The useful thing about futures, as opposed to various IVar packages, is > handling the case where the forked operation ends with an exception. > The exception becomes the return value of the promise. The "get" > operation rethrows it, the "wait" operation returns it as (Left ...). > > There is also an "abort" command to kill a promise. The dead promise > may then have an exceptions as its value. > > The "plus a bit more" than C++ is the nonblocking "addTodo" feature. > This takes a continuation function from the "Either SomeException a" to > an IO operation. These continuation functions get queued and they are > run immediately after the the forked operation completes. Once > completed any new "addTodo" continuations run immediately. > > These continuations allow you to race a list of action and take the > first one done, or to collect the answers as they complete into a Chan. > Both of those options are demonstrated in Future.hs as racePromises and > forkPromises. > > It should be safe to use "unsafePerformIO . get" or "unsafePeformIO . > wait" to get lazy access to the result, which is itself immutable once set. > > Cheers, > Chris I used something similar to this for a monadic opengl "RPC" mechanism. The need was for transparently creating OGL objects from any context. Returning thunks that block on the result when finally evaluated works out real nice in practice. Its easy to manage, and performs better than dealing with several OS threads and GL contexts.

1 0

I want to write a compiler
by Loup Vaillant 09 Mar '09

09 Mar '09

This is a homework question. I am mainly looking for guidance or pointers. Source code, even. (Not GHC's, though, it is too complicated for me right now. The AST of STG is fine, but the rest kinda scares me.) Ideally, I would very much like to compile to C. The requirements are easily stated. My language must -> be lazy by default, -> support algebraic data types and case expressions (unless I can get away with encoding them as functions), -> have some kind of FFI with C (I suppose it imply some support for unboxed values). There is also the first class functions thing, but lambda lifting is okay. I don't care about any kind of execution efficiency. I just want my compiler to be As Simple As Possible. Eventually, I intend to bootstrap. Then, I will start dreaming about doing better than the Mighty Simons. (I can dream, right?) I figured I would start by the back-end. I began to write 15 pathetic lines of code to start an STG machine. Needles to say, I am stuck. I don't know where to start. I am already familiar with some papers. In particular, [1] (on template instantiation), [2] and [3]. I once wrote a simple graph reducer using template instantiation (about 20 lines at most). I have chosen the STG machine because I thought i would be easier to get an FFI and case exprs out of it. Maybe I am wrong, and template instantiation is really easier. There is also the BRISK machine, and GRIN. But the paper I read were less readable to me than the three mentioned. So, If any of you can give me some pointer or guidance or advice about where to start, I would be very grateful. Loup [1] http://www.cs.otago.ac.nz/cosc459/books/pjlester.pdf [2] http://research.microsoft.com/~simonpj/Papers/spineless-tagless-gmachine.ps… [3] http://research.microsoft.com/~simonpj/Papers/eval-apply/index.htm

9 12

Minimum Int?
by Colin Paul Adams 09 Mar '09

09 Mar '09

Is there a function that yields the minimum value of Int on an implementation? -- Colin Adams Preston Lancashire

3 2

Partial pattern matching
by Peter Verswyvelen 09 Mar '09

09 Mar '09

In Haskell, a data constructor can be used partially applied: data Pair a b = P a b f = P 1 however, I cannot do "partial pattern matching", e.g firstCoord (P x) = x does not work. I guess a very important reason must exist why this is the case?

7 8

Parsing floating point numbers
by Bjorn Buckwalter 09 Mar '09

09 Mar '09

Hi all, What is your preferred method of parsing floating point numbers (from String to Float/Double)? Parsec it seems only does positive floats out of the box and PolyParse requires the float to be on scientific form (exponential). While I've worked around these shortcomings in the past I feel that I am reinventing the wheel as surely I am not the only to run into these limitations. How do you parse your floats? Can you recommend a parsing library that handles them solidly? (For my current needs the formats accepted by "read" are sufficient, but I want reasonable error handling (Maybe or Either) instead of an exception on bad inputs.) Thanks, Bjorn Buckwalter

7 6