March 2007 - Haskell - Haskell.org

CFP: HLPP 2007 (Tokyo, July 23-24, 2007)
by Zhenjiang Hu 09 Mar '07

09 Mar '07

We apologize if you receive multiple copies. --------------------------------------------------------------- Call For Papers Fourth International Workshop on HIGH-LEVEL PARALLEL PROGRAMMING AND APPLICATIONS (HLPP 2007) The University of Tokyo, Tokyo, Japan July 23-24, 2007 http://f.loulergue.free.fr/HLPP/hlpp2007/ --------------------------------------------------------------- AIMS AND SCOPE Bill McColl's post Sequential Computing Considered Harmful is an excellent summary of today's situation (http://www.computingatscale.com) Sequential computing cannot go further. Major companies in the computing industry now recognizes the urgency of reorienting an entire industry towards massively parallel computing. The trend is towards the increase of cores in processors and the need for scalable computing everywhere. But parallel and distributed programming is still dominated by low-level techniques such as send/receive message passing. Thus high-level approaches should play a key role in the shift to scalable computing in every computer. This workshop follows the first three HLPP workshops held in 2001, 2003 and 2005. It is aimed at: - computer science researchers, practitioners, graduate students - scientific computing researchers, practitioners, graduate students - high-performance application developers (e.g. in DBMS, data-mining, parallel model checking, virtual reality) TOPICS We welcome submission of original, unpublished papers in English on topics including (but not limited to) the following aspects of multi-core, concurrent, parallel, distributed, meta and grid computing: - high-level algorithms for parallel, communication-efficient and external-memory computation - high-level programming models (BSP, CGM, LogP, MPM, etc.) and tools - high-level programming models, tools and algorithms for multi-threaded, reconfigurable and multi-core computing - parallel programming methodologies - algorithmic skeletons and constructive methods - performance models and performance evaluation - high level resource-aware approaches object, functional, logic, constraint programming (semantics and implementation) - verification of parallel and distributed programs - security in high-level approaches - applications using high-level languages and tools (scientific computing, DBMS, model-checking, virtual reality, ...) - teaching experience with high-level tools and methods Accepted papers should be presented at the workshop. Revised versions will be published in a special issue of an international journal (previous HLPP workshops were published in Parallel Processing Letters, World Scientific Publishing) provided revisions suggested by the referees are made. IMPORTANT DATES 15 April 2007: Full paper due 28 May 2007: Notification 2 July 2007: Camera-ready paper due 23-24 July 2007: Workshop 17 September 2007: Journal version due PROGRAMME COMMITTEE Manuel Chrakravarty (Univ. of New South Wales, Australia) Murray Cole (Univ. of Edinburgh, UK) Alexandros Gerbessiotis (NJIT, USA) Sergei Gorlatch (Univ. of Muenster, Germany) Zhenjiang Hu (Univ. of Tokyo, Japan) Christoph Kessler (Linkopings Universitet, Sweden) Herbert Kuchen (Univ. of Muenster, Germany) Rita Loogen (Univ. of Marburg, Germany) Frederic Loulergue (Univ. of Orleans, France) Quentin Miller (Somerville College, Oxford, UK) Susanna Pelagatti (Univ. of Pisa, Italy) Alexander Tiskin (Univ. of Warwick, UK) Kazunori Ueda (Waseda University, Japan) CHAIRS AND ORGANIZERS Zhenjiang HU Information Processing Laboratory Department of Mathematical Informatics Department of Mathematical Engineering and Information Physics The University of Tokyo Frederic LOULERGUE Laboratoire d'Informatique Fondamentale d'Orleans Faculty of Sciences Computer Science Department University of Orleans PAST HLPP WORKSHOPS HLPP 2005 was held in Coventry (July 4-5, 2005). HLPP 2003 was held in Paris (June 16-18, 2003). Parallel Processing Letters (Volume 13, issue 3) contains 14 revised papers presented at the workshop. HLPP 2001 was held in Orleans (March 26-27, 2001). Parallel Processing Letters (Volume 11,issue 4) contains 8 revised papers presented at the workshop.

1 0

Summer of Code questions
by Wolfgang Jeltsch 09 Mar '07

09 Mar '07

Hello, I’ve some questions regarding the Haskell.org involvement in the Google Summer of Code. First, what organization is “Haskell.org”? Is this a real organization, i.e., a legal entity? Does Haskell.org get the $ 500 for each successful project? If yes, what does Haskell.org do with this money? Furthermore, the Google Summer of Code FAQ talks about *the* project of a organization which seems to indicate that each organization has only one SoC project. What is the project of the Haskell.org organization then? Is this the general project “The Haskell language plus associated libraries and tools”? The FAQ says “Mentor organizations must be organizations or individuals running an active and viable open source or free software project”. Does this mean that the above-mentioned general project has to be active and viable (which it is) or that each concrete idea has to be part of an active and viable project? That is, is it allowed to start a new concrete project by letting a student code for it as part of the SoC or wouldn’t this be possible since a new project isn’t yet active and viable? Who decides which ideas will be worked on and who decides which student works on which? What are the obligations of a mentor? Is it okay to create an idea, become the mentor for this idea and propose a concrete student for it? What do I have to do to become a mentor? What are Haskell.org’s criteria for selecting mentors? According to the above-mentioned FAQ, Google wants to know these criteria as specifically as possible. Finally, should the ideas chosen so that they are solvable by a student working full time on them during the three months of the SoC? The problem is that at many German universities lectures go until July. At our university, for example, lecture time ends in mid of July and is followed by two weeks for exams. So a faithful student would only have the last month for working full time on the SoC. Thank you very much for any clarifications. Best wishes, Wolfgang

5 5

ANN: binary arithmetic type library over natural kinds
by oleg＠pobox.com 09 Mar '07

09 Mar '07

Chung-chieh Shan and I would like to announce the type-level Haskell library for arbitrary precision binary arithmetic over natural kinds. The library supports addition/subtraction, predecessor/successor, multiplication/division, exp2, full comparisons, GCD, and the maximum. At the core of the library are multi-mode ternary relations Add and Mul where _any_ two arguments determine the third. Such relations are especially suitable for specifying static arithmetic constraints on computations. The type-level numerals have no run-time representation; correspondingly, all arithmetic operations are done at compile time and have no effect on run-time. http://pobox.com/~oleg/ftp/Computation/resource-aware-prog/BinaryNumber.hs We used the arithmetic type library to statically enforce validity, range, size, and alignment constraints of raw memory pointers, and to statically enforce protocol and time-related constraints when accessing device registers. The following near final full paper http://pobox.com/~oleg/ftp/Computation/resource-aware-prog/tfp.pdf describes the arithmetic type library, type-level records, type-level programming with regular Haskell terms, and two sample applications. We will greatly appreciate all the comments (especially those received before noon on Friday).

3 2

Re: ANN: binary arithmetic type library over natural kinds
by oleg＠pobox.com 09 Mar '07

09 Mar '07

David Roundy wrote: > data Arr a n = ... -- n is a type-level variable describing the dimension > -- a is the data type stored > withArrayFromList :: [a] -> (Nat n => Arr a n -> b) -> b > ... > But what I'm dreaming of is that the type system would enforce those > run-time checks when necesary. e.g. > > catArray :: Add n1 n2 n3 => Arr a n1 -> Arr a n2 -> Arr a n3 > > I imagine this is harder, but it's needed if these type-level binaries > will be useful for the sorts of programs I tend to write--which have > run-time sizes. I believe the lightweight static capabilities approach will be more appropriate here. It already provides the function withArrayFromList, which is called `brand'. For example, KMP-deptype.hs includes > -- brand the packed string > brandPS:: PackedString > -> (forall r. (BPackedString r, BIndexP1 r) -> w) -> w -> w which is quite like `withArrayFromList' only using `r' rather that 'n' and with an extra argument which is returned when the source packed string turns out empty. The approach almost fulfills your dream: only, instead of the `system' inserting run-time checks, it is the programmer who inserts the checks, to please the typechecker. In the catArray example, one would write something like that > -- kernel > data Arr a = ... > newtype BArr n a = BArr{unBArr:: Arr a} -- BArr data ctor not exported > brand :: Arr a -> (forall n. (BArr n a, BIx n) -> w) -> w > -- User code > catArray :: BArr n1 a -> BArr n2 a -> Arr a > catArray a1 a2 = concatArr (unBArr a1) (unBArr a2) > test a1 a2 = brand (catArray a1 a2) (\newarray itsbound -> ...) The user can easily write catArray to create an unbranded array. One needs branding however to be able to index into the array: thus one has to invoke brand (the `smart' constructor of the BArr type), which will do the run-time check. The kernel may provide other constructors of BArr from arguments of some particular types, which let us avoid the run-time check. Alas, the style of writing is a bit like CPS -- OTH, we write monadic code, which is CPS in thin disguise, with no complaints... One can use existentials to make the code look more direct-style, however. The programming becomes a bit like theorem proving: the indexing operation needs BArr, so we have to consider all the ways we can get a value of that type. The kernel provides a few constructors; one of which, brand, always applies, but it involves a run-time check. There may be other constructors; but they require arguments of particular types. We should see if we can produce the values of those types from already available values, etc. This is back-chaining theorem proving.

1 0

method annotations
by John Meacham 08 Mar '07

08 Mar '07

On Sat, Mar 03, 2007 at 08:23:08AM -0500, Isaac Dupree wrote: > Unfortunately it would be difficult for the compiler to know that no-one > is going to make an instance that does depend on its argument (as long > as it does not do full-program analysis to determine the layout of data > structures). Maybe there should be some sort of annotation on class > method arguments that are supposed to be ignored, even if only a pragma, > if it would help compilers figure this out. indeed, I have given this some thought too, but not just for whether a type will be unused. my current thinking is something like > class Bits a where > shift :: a -> !Int -> a > > class Eq a where > (==) :: a -> a -> !Bool > > class Typeable a where > getType :: ~a -> TypeRep where '!Int' means that the class is strict in the Int argument. and ~ means it is unused. note, these are not just hints, but actually affect the class translation. shift as above would translate to something ilke > shift :: BitsDict a -> Int -> a > shift bd x i = i `seq` shiftInstance bd x i where shiftInstance selects the appropriate value from the dictionary of course, the real benefit is now that we can do unboxing on class methods! we can treat the above class declaration as if it were class Bits a where shift :: a -> Int# -> a safely and rewrite instance declarations appropriately likewise, the ! on the return value in Eq means that we should unbox it. (this is actually always safe, so perhaps a pragma will do for this, "UNPACK" seems appropriate) for the other case > class Typeable a where > getType :: ~a -> TypeRep would create a method like so > getType :: TypeableDict a -> a -> TypeRep > getType td a = getTypeInstance td (undefined `asTypeOf` a) which causes getType to always throw away its argument, allowing the compiler to make assumptions based on it. another implicit optimization that is going on in my translations above is eta-expansion of the class methods (which is independent of the optimizations given above). this is potentially unsafe, but I have been performing the optimization in jhc without ill effect and with noticable benefit. however, you can specify the "NOETA" pragma in order to suppress this auto-eta-expansion as it does decrease sharing. another option for controlling eta expansion would be something like class Eq a where (==) :: a -> !(a -> Bool) which makes explicit that the partial result of (==) applied to one argument could usefully be shared. currently, these optimization are availabe in jhc, but only via pragmas and some built in heuristics/special knowledge in the compiler. Implementing the above syntax has been on my todo list. Anecdotal examination of core leads me to believe they could be signifigantly beneficial. John -- John Meacham - ⑆repetae.net⑆john⑈

1 0

Google summer of code
by Malcolm Wallace 07 Mar '07

07 Mar '07

Google Summer of Code As many of you will already know, Google is running its "Summer of Code" project again this year, and haskell.org is once again going to apply to be a mentoring organisation. Are you a student who would like to earn money for hacking in Haskell? Or are you a non-student who has a cool idea for a coding project but no time to do it yourself? Well, our wiki to gather ideas is now up-and-running again: http://hackage.haskell.org/trac/summer-of-code Add yourself to the list of interested people! There are some ideas still there from last year. Add new ones! Google will start accepting student applications in just over a week's time, but now is the time to start using our own wiki to match up interesting ideas with interested people. The official timeline is as follows: March 12: Mentoring organization application deadline March 14: List of accepted mentoring organizations published March 14: Student application period opens March 24: Student application deadline Interim Period: we review and rank student proposals April 9: List of accepted student applications published Interim Period: Students learn more about their project communities May 28: Students begin coding; Google begins issuing initial payments July 9: Students upload code to code.google.com/hosting; July 16: Google begins issuing mid-term payments August 20: Students upload code to code.google.com/hosting; August 31: Final evaluation deadline; Google begins issuing final payments Regards, Malcolm

3 2

Fwd: HFL07, student stipend(s) available
by Koen Claessen 07 Mar '07

07 Mar '07

Workshop on Hardware design and Functional Languages HFL'07, Braga, Portugal, March 24 and 25 (with the ETAPS conferences) Student stipend(s) available, deadline March 13 HFL07 is pleased to announce the availability of one or two stipends for students to attend the workshop. We will give preference to undergraduate or masters students, but will also consider PhD students. The stipend should cover all or part of registration at the workshop, the workshop dinner on sunday night, travel and student accommodation. The student should write a letter to the workshop chairs (hfl07(a)hflworkshop.org) outlining their costs, other financial support that they have obtained (e.g. from their own university) and why they would like to attend the workshop. They should also include a statement from a teacher or supervisor supporting their case (and including an email address). Deadline for receipt of this application is midnight on tuesday March 13th, UK time. For information on the workshop programme, see http://www.hflworkshop.org/ Information about the ETAPS group of conferences, registration, student rooms etc. is at http://www.di.uminho.pt/etaps07/ Please encourage a good undergrad. or masters student to apply for a stipend and come to the workshop! with best wishes Andy Martin, Carl-Johan Seger and Mary Sheeran

1 0

Re: [Haskell] Haskell Dependency/Installation of packages (FilePath/MissingPy)
by Gottfried F. Zojer 06 Mar '07

06 Mar '07

> -----Ursprüngliche Nachricht----- > Von: "Neil Mitchell" <ndmitchell(a)gmail.com> > Gesendet: 06.03.07 10:33:11 > An: "Gottfried F. Zojer" <goenzoy(a)web.de> > CC: haskell(a)haskell.org > Betreff: Re: [Haskell] Haskell Dependency/Installation FilePath > Hi > > > Loading package base ... linking ... done. > > [1 of 1] Compiling Main ( C:/ghc/ghc-6.6/filepath-0.11/filepath/Setu > > p.hs, interpreted ) > > Ok, modules loaded: Main. > > *Main> > > You have probably downloaded FilePath, but not installed it. Try: > > cd C:/ghc/ghc-6.6/filepath-0.11/filepath > runhaskell Setup configure > runhaskell Setup build > runhaskell Setup install > > Then attempt the installation of MissingH again. > > Thanks > > Neil > Hallo Neil , Thanks for the advice.Well Filepath was installed and at the end everything looked fine. MissingH installation with (winbuild) was going through.I m further now with MissingPy and its the same symptom.(see compile log) Like on the first attempt with MissingH !? [44 of 46] Compiling System.Path.Glob ( src/System/Path/Glob.hs, dist\build/Syst em/Path/Glob.o ) [45 of 46] Compiling System.Debian.ControlParser ( src/System/Debian/ControlPars er.hs, dist\build/System/Debian/ControlParser.o ) [46 of 46] Compiling System.Debian ( src/System/Debian.hs, dist\build/System/ Debian.o ) C:\ghc\ghc-6.6\bin\ar.exe: creating dist\build\libHSMissingH-0.18.0.a C:\server\missingh_0.18.1\missingh> C:\Python24\missingpy_0.1.0\missingpy>runhaskell Setup configure Setup: Warning: Unknown field 'options-ghc' Setup: Warning: Unknown field 'hidden-modules' Setup: Warning: Unknown field 'extra-libs' Setup: Warning: No license-file field. Configuring MissingPy-0.1.0... configure: Dependency haskell-src-any: using haskell-src-1.0 Setup: cannot satisfy dependency MissingH>=0.9.0 <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<< C:\Python24\missingpy_0.1.0\missingpy>runhaskell Setup build Setup: Warning: Unknown field 'options-ghc' Setup: Warning: Unknown field 'hidden-modules' Setup: Warning: Unknown field 'extra-libs' Setup: error reading ./.setup-config; run "setup configure" command? C:\Python24\missingpy_0.1.0\missingpy> C:\Python24\missingpy_0.1.0\missingpy>python gencabal.py *** Generating MissingPy.cabal based on these settings *** Please edit MissingPy.cabal if the detected settings are *** incorrect. Include path for Python headers: C:\Python24\include Library paths for Python library: None, C:\Python24\Lib\site-packages Python library name: python2.4 Thanks for any feedback Best regards Gottfried www.wirtschaftswunder.co.uk www.5152.eu

1 0

Haddock not being able to read interface file
by Wolfgang Jeltsch 06 Mar '07

06 Mar '07

Hello, I want to generate Haddock documentation for some Haskell files and want to create hyperlinks to the base package’s documentation. I’ve downloaded <http://www.haskell.org/ghc/docs/latest/html/libraries/base/base.haddock> and use the option --read-interface=http://www.haskell.org/ghc/docs/latest/html/libraries/base,base.haddock in my Haddock command line. However, Haddock outputs the following: haddock: Warning: The interface file "base.haddock" could not be read. Maybe it's from a later version of Haddock? This happens with Haddock 0.8 as well as the current development version of Haddock. What’s wrong here? Best wishes, Wolfgang

2 2

Haskell Dependency/Installation FilePath
by Gottfried F. Zojer 06 Mar '07

06 Mar '07

Hallo to all, Tried to install MissingH ,something what failed because Filepath is missing. Any idea why the setup file isnt in the position to find it,even if it is there(see GHCi log down ) Any feedback welcome Best regards Gottfried www.wirtschaftswunder.co.uk www.5152.eu ___________________________________________________________________________________ missingh>setup configure setup: Warning: No license-file field. Configuring MissingH-0.18.0... configure: C:\ghc\ghc-6.6\bin\ghc-pkg.exe configure: Dependency network-any: using network-2.0 configure: Dependency parsec-any: using parsec-2.0 configure: Dependency base-any: using base-2.0 configure: Dependency haskell98-any: using haskell98-1.0 configure: Dependency mtl-any: using mtl-1.0 configure: Dependency HUnit-any: using HUnit-1.1 configure: Dependency regex-compat-any: using regex-compat-0.71 configure: Dependency QuickCheck-any: using QuickCheck-1.0 setup: cannot satisfy dependency FilePath-any C:\server\missingh_0.18.1\missingh>setup build setup: error reading ./.setup-config; run "setup configure" command? ___ ___ _ / _ \ /\ /\/ __(_) / /_\// /_/ / / | | GHC Interactive, version 6.6, for Haskell 98. / /_\\/ __ / /___| | http://www.haskell.org/ghc/ \____/\/ /_/\____/|_| Type :? for help. Loading package base ... linking ... done. [1 of 1] Compiling Main ( C:/ghc/ghc-6.6/filepath-0.11/filepath/Setu p.hs, interpreted ) Ok, modules loaded: Main. *Main>

2 1