- Haskell-Cafe - Haskell.org

ANN: New release of ZeroTH
by Robin Green 23 Jun '09

23 Jun '09

This announcement will only be of interest to Haskell programmers using, or thinking of using, Template Haskell. I am very pleased to announce a new release (2009.6.23.3) of ZeroTH (also known as zeroth), a tool for preprocessing Haskell code to run splices and remove Template Haskell dependencies: http://hackage.haskell.org/package/zeroth For example, you could use it to reduce the size of your Haskell program binaries, if you use Template Haskell. (zeroth now actually uses itself on itself, for this very purpose, and you can use the Distribution.ZeroTH module for that in your own cabal projects.) It's still very much a hack - in that it would probably be more efficient and clean to implement it as an option to GHC - but hopefully now at least it's a slightly *better* hack. Lemmih has kindly agreed to hand over the job of being maintainer of ZeroTH to me. So please send any patches or complaints my way. The new darcs repository for the project, which is a darcs 2 repository, is at: http://patch-tag.com/publicrepos/zerothDarcs2 For this release, the "cabal" command-line tool is the only officially supported means of installation. So you should do something like: cabal update cabal install zeroth to install it. To get a summary of command-line options, type: zeroth --help Major changes from previous releases: * Much more Haskell code is now supported. This is partly because the code has been updated to work with the just-released haskell-src-exts 1.0.0; and partly because it supports a wider variety of Template Haskell code, including code which refers to definitions declared in the same source file. * It is now compulsory to specify any Haskell language extensions you are using in the input file, with a LANGUAGE pragma. (For debugging purposes, zeroth prints to stderr the result of parsing the LANGUAGE pragma, if any. If haskell-src-exts doesn't understand the LANGUAGE pragma, this will be Nothing. But that shouldn't happen.) * Template Haskell imports are now *automatically* stripped by default by zeroth, and so cpphs doesn't need to be run if you aren't otherwise using CPP. * New command-line options have been added (see zeroth --help). * Error messages are now more informative. * Librification! Almost anything can benefit from being made into a library, and zeroth is no exception. Of course, the executable is still there. * You can still use it on certain code which doesn't compile in GHC (yes, really, I've tested this!) but you now have to use the --only-splices option if you want to do this. Obviously, the splices themselves have to compile. --only-splices also makes zeroth faster. It is very lightly tested, so it might not work for your code out of the box. Please let me know if you have any problems with it, no matter how small. Happy hacking, -- Robin

1 0

Re: [Haskell] ANN: Reusable Corecursive Queues via Continuations
by Felipe Lessa 23 Jun '09

23 Jun '09

On Mon, Jun 22, 2009 at 10:31:34PM -0400, Leon Smith wrote: > Also, a substantially revised draft of the associated paper, "Lloyd > Allison's Corecursive Queues: Why Continuations Matter" is now available. > [3] This paper will appear in the upcoming Monad Reader issue 14, and > comments would be most appreciated. As per [1], it seems n+k patterns will be removed from Haskell', so you may want to remove them from the paper as well. [1] http://hackage.haskell.org/trac/haskell-prime/wiki/Status -- Felipe.

1 0

memoization using unsafePerformIO
by Jan Christiansen 23 Jun '09

23 Jun '09

Hi, I have tried to implement a memo function using stable names and weak pointers like it is presented in the paper "stretching the storage manager". There is an abstract datatype SNMap a b which implements a map that maps values of type StableName a to values of type b. The map is located in an IORef. {-# NOINLINE memo #-} memo :: (a -> b) -> a -> b memo f = unsafePerformIO (do tref <- newSNMap return (applyWeak f tref)) {-# NOINLINE applyWeak #-} applyWeak :: (a -> b) -> SNMap a (Weak b) -> a -> b applyWeak f tbl arg = unsafePerformIO (do sn <- makeStableName arg lkp <- lookupSNMap tbl sn case lkp of Nothing -> not_found sn Just weak -> do val <- deRefWeak weak case val of Just result -> return result Nothing -> not_found sn) where not_found sn = do let res = f arg weak <- mkWeak arg res Nothing insertSNMap tbl sn weak return res Using this memo function I have implemented a fibonacci function like it is defined in the paper. fib :: Int -> Int fib = memo ufib ufib :: Int -> Int ufib 0 = 1 ufib 1 = 1 ufib n = fib (n-1) + fib (n-2) When I compile this program with 6.10.3 fib yields the correct results up to 18 but for 19 it detects a loop. Can anybody give me a hint how I can avoid this behaviour (and still use unsafePerformIO ; ) )? Cheers, Jan

1 0

ANN: haskell-src-exts-1.0.0
by Niklas Broberg 23 Jun '09

23 Jun '09

Fellow Haskelleers, It is with great pleasure I hereby announce the first stable release of the haskell-src-exts package, version 1.0.0! haskell-src-exts is a package for Haskell source code manipulation. In particular it defines an abstract syntax tree representation, and a parser and pretty-printer to convert between this representation and String. It handles (almost(*)) all syntactic extensions to the Haskell 98 standard implemented by GHC, and the parsing can be parametrised on what extensions to recognise. I wish to thank the glorious Haskell community, for giving me the chance to work on this project as part of Haskell.org's GSoC programme, but also for simply being such a nice place to be! Special thanks to everyone who helped me with testing and bug reports during the final stretch of release candidates, in particular the many excellent reports from Ganesh Sittampalam, Sebastian Fischer, and Neil Mitchell who is also mentoring the project. You're all awesome! haskell-src-exts-1.0.0: ================= Via cabal: cabal install haskell-src-exts-1.0.0 Via darcs: darcs get http://code.haskell.org/haskell-src-exts On hackage: http://hackage.haskell.org/package/haskell-src-exts-1.0.0 Changes from 0.5.7, the last release candidate: ====================================== * CPP lines are no longer ignored, which means haskell-src-exts will now invariably give a parse error on files with CPP pragmas in them. CPP is not supported by haskell-src-exts, and this is more intuitive than parsing e.g. all branches of an #if pragma, which is invariably give unintended results. * Fixed a stupid introduced bug where extensions passed int he parse mode were ignored. * fromParseOk is now exported, not just defined (doh). * ScopedTypeVariables now implies TypeOperators, as per GHC. I'm sure there are more implications that are missing from haskell-src-exts, I will add them as I find out about them. Thanks once again, and Happy Haskell Hacking to all! Cheers, /Niklas (*) The only two exceptions are NewQualifiedOperators and UnicodeSyntax.

2 2

Ann: Emping-0.6 and Tests/Examples
by Hans van Thiel 23 Jun '09

23 Jun '09

Hello All, Version 0.6 of Emping is now, imo, at the level of a working prototype. It is available on Hackage (but won't build there because it uses Gtk2Hs. It did build with Cabal on FC8 with GhC 6.8 and Gtk2Hs 0.9.13). The package includes a user guide, but see the update on my web site about two UI corner case bugs (program hangs). My site http://www.muitovar.com discusses three examples of possible use, with tests using SQLite, OO Calc and dotty. There's also a .zip archive of all sources (two data sets from the UCI machine learning repository) and .csv and .dot files with the results. All comments most welcome... Emping is a (prototype) interactive tool for the discovery and analysis of (universal, not statistical) predictive rules in tables of nominal data. Best Regards, Hans van Thiel

1 0

INAP 2009: 2nd Call for Papers
by Pedro Salgueiro 23 Jun '09

23 Jun '09

[apologies for cross-posting; please distribute] --- (PLEASE DISTRIBUTE) ------------------------------------------------------------ Second Call for Papers INAP 2009 18th International Conference on Applications of Declarative Programming and Knowledge Management November 5-7, 2009 Evora, Portugal http://www.di.uevora.pt/inap2009/ http://inap.dialogengines.com/ ------------------------------------------------------------ Organized by the Portuguese AI Society (APPIA), the INAP Committee and the Society of Logic Programming (GLP e.V.) == Overview == Declarative Programming is a family of advanced paradigms for the modeling and solving of complex problems. These specification and implementation methods have attracted more and more attention over the past years, e.g. in the domains of databases and natural language processing, for modeling and the processing of combinatorial problems, and for establishing systems for the web. == INAP 2009 == INAP is a communicative and dense forum for intensive discussion of applications of important technologies related to Prolog, Logic and Constraint Programming as well as closely related advanced software. It comprehensively covers the impact of programmable logic solvers in the Internet Society, its underlying technologies, and leading edge applications in industry, commerce, government, and societal services. INAP 2009 continues a tradition of successful workshops cast around the applications of declarative programming, which were held in Kobe (1997), Tokyo (1995, 1996, 1998 - 2001), Potsdam (2004), Fukuoka (2005) and Wuerzburg (2007). We invite the submission of high quality papers on the described topics, especially, but not exclusively, on different aspects of Declarative Programming, Constraint Processing and Knowledge Management as well as their use for Distributed Systems and the Web: - Knowledge Management, e.g. Data Mining, Decision Support, Deductive Databases - Distributed Systems and the Web, e.g. Agents and Concurrent Engineering, Semantic Web - Constraints, e.g. Constraint Systems, Extensions of Constraint (Logic) Programming - Theoretical Foundations, e.g. Deductive Databases, Nonmonotonic Reasoning - Systems and Tools for Academic and Industrial Use - Knowledge-based Web Services - Logic Solvers and Applications == Workshop Format == The technical program of the workshop will include invited presentations (to be announced), regular technical sessions with presentations of the accepted papers, system demonstrations and a panel discussion. == Conference Venue == The conference will be held at the University of Evora, Portugal in November 5-7, 2009. Evora is a nice and quiet historical city located in the south of Portugal that can be reached from Lisbon by train or coach in under 2 hours. It is a small city of 60.000 inhabitants, 120 km inland from Lisbon and classified by Unesco as World Heritage. The University of Evora was established in the 16th Century and is the 2nd oldest Portuguese University. The social program is promising since the region is very rich in historical sites (Stone Age, Roman, Medieval and Renaissance remains) and also offers a very special gastronomy. See http://en.wikipedia.org/wiki/%C3%89vora for more information. == Important Dates == Paper Submission Deadline: June 29, 2009 Notifications to Authors: August 17, 2009 Camera-ready Version Deadline: September 14, 2009 INAP 2009 Workshop: November 5-7, 2009 == Submission Guidelines == Participants should submit a paper (maximum 15 pages, PDF format), describing their work in topics relevant to the workshop. Accepted papers will be presented during the workshop. At least one author of an accepted contribution is expected to register for the workshop, and present the paper. All submissions should include the author's name(s), affiliation, complete mailing address, and email address. Authors are requested to prepare their submissions, following the LNCS/LNAI Springer format. Please see: http://www.springer.de/comp/lncs/authors.html for further details. The submission should be submitted through the electronic submission site: http://www.easychair.org/conferences/?conf=inap2009 The deadline for receipt of submissions is June 29, 2009. Papers received after this date will not be reviewed. Eligible papers will be peer-reviewed by at least three members of the Program Committee. Authors will be notified via email of the results by August 17, 2009. Authors of accepted papers are expected to improve their paper based on reviewers' comments and to send a camera ready version of their manuscripts by September 14, 2009. Accepted papers will be included in the workshop proceedings, which will be distributed to the participants. As in previous editions, we plan to publish selected papers in a proceedings volume in the Springer Lecture Notes in Artificial Intelligence (LNAI) series. == Organizing Committee == Vitor Nogueira vbn AT di.uevora.pt Salvador Abreu spa AT di.uevora.pt Pedro Salgueiro pds AT di.uevora.pt Universidade de Evora Portugal == Program Committee == Salvador Abreu, University of Evora, Portugal (co-chair) Sergio Alvarez, Boston College, USA Philippe Codognet, CNRS/JFLI, Tokyo, Japan Vitor Santos Costa, University of Porto, Portugal Daniel Diaz, University of Paris I, France Ulrich Geske, University of Potsdam, Germany Gopal Gupta, UT Dallas, USA Petra Hofstedt, Technical University of Berlin, Germany Ulrich Neumerkel, Technical University of Vienna, Austria Vitor Nogueira, University of Evora, Portugal Enrico Pontelli, New Mexico State University, USA Irene Rodrigues, University of Evora, Portugal Carolina Ruiz, Worcester Polytechnic Institute, USA Dietmar Seipel, University of Wuerzburg, Germany (co-chair) Terrance Swift, CENTRIA, Portugal Hans Tompits, Technical University of Vienna, Austria Masanobu Umeda, Kyushu Institute of Technology, Japan Armin Wolf, Fraunhofer FIRST, Berlin, Germany Osamu Yoshie, Waseda University, Japan == Contact Information == inap2009(a)di.uevora.pt Universidade de Evora Departamento de Informatica Largo dos Colegiais, 2 7004-516 Evora - PORTUGAL

1 0

IMO ghc 6.8.2 misleading warning messages
by Vasili I. Galchin 23 Jun '09

23 Jun '09

Hello, For: parseOptional = Parse.parseOptional I got warning messages: Swish/HaskellUtils/ParseURI.hs:77:4: Warning: Defined but not used: `parseOptional' Swish/HaskellUtils/ParseURI.hs:77:4: Warning: Definition but no type signature for `parseOptional' Inferred type: parseOptional :: forall a b. Parse.Parser a [b] -> [a] -> [([b], [a])] IMO the warning messages should be reversed! (parseOptional was never "used" .. called ... but I would strongly prefer the type signature warning first. Kind regards, Vasili

2 1

Getting my mind around UArray -> STUArray conversion
by Scott Michel 23 Jun '09

23 Jun '09

I'm trying to get my mind around how to thaw and then freeze a UArray. Theoretically, what I've written below should be a no-op, but I keep getting typing errors that I can't figure out. GHCI 6.10.3 says: Couldn't match expected type `UArray ix a' against inferred type `ST s (STUArray s ix1 e)' In the first argument of `(>>=)', namely `(unsafeThaw mem :: ST s (STUArray s ix e))' In the expression: (unsafeThaw mem :: ST s (STUArray s ix e)) >>= (\ mmem -> unsafeFreeze mmem) In the definition of `wombat': wombat val idx mem = (unsafeThaw mem :: ST s (STUArray s ix e)) >>= (\ mmem -> unsafeFreeze mmem) I'm figuring that usafeThaw with the type annotation should have given GHIC enough clue. Any suggestions? -scooter (WOMBAT = Waste Of Money Brains And Time) import Control.Monad; import Control.Monad.ST; import Data.Array.ST; import Data.Array.Unboxed; import Data.Array.MArray; import Data.Word; wombat :: (IArray UArray e, Ix ix, MArray (STUArray s) e (ST s)) => e -> ix -> UArray ix e -> UArray ix e wombat val idx mem = (unsafeThaw mem :: ST s (STUArray s ix e)) >>= (\mmem -> unsafeFreeze mmem)

6 7

STM/Data Invariant related Segfault with GHC 6.10.3
by Jan Schaumlöffel 23 Jun '09

23 Jun '09

Hello, I just discovered that programs compiled with GHC 6.10.3 segfault when accessing a TVar created under certain conditions. This happens when the TVar is created and a data invariant is added (using alwaysSucceeds) in the same atomic block. The invariant does not necessarily have to read the TVar in question, a mere "alwaysSucceeds (return ())" is enough. It looks like by adding the invariant the TVar allocation is not kept after the block. The appended code snippets show how to reproduce this behaviour. IMHO the first version should work, too. Is there anything one can do to (reliably) work around this problem? Is there maybe an underlying problem that causes this? Regards, Jan module Main where import GHC.Conc -- this segfaults: main = do { t <- atomically (do { t1 <- newTVar 0 ; alwaysSucceeds (readTVar t1) ; return t1 }) ; atomically $ readTVar t } -- it works if written like this: main2 = do { t <- newTVarIO 0 ; atomically $ alwaysSucceeds (readTVar t) ; atomically $ readTVar t } -- works also if no invariant is added: main3 = do { t <- atomically (do { t1 <- newTVar 0 ; return t1 }) ; atomically $ readTVar t } -- works also if invariant is added later: main4 = do { t <- atomically (do { t1 <- newTVar 0 ; return t1 }) ; atomically $ alwaysSucceeds (readTVar t) ; atomically $ readTVar t } -- If you're happy and you know it, syntax error!

2 1

haskelldb + sqlite problem.
by Magicloud Magiclouds 22 Jun '09

22 Jun '09

Hi, I am using haskelldb and haskelldb-hdbc-sqlite3. Well, I finally got the source compiled and ran, I got this error: App: user error (SQL error: SqlError {seState = "", seNativeError = 21, seErrorMsg = "prepare 74: SELECT subject,\n timestamp\nFROM notes as T1\nORDER BY timestamp DESC: library routine called out of sequence"}) Any clue what I should check? Thanks. -- 竹密岂妨流水过山高哪阻野云飞

3 6