August 2009 - Haskell-Cafe

Installing documentation with cabal
by Maciej Piechotka 18 Aug '09

18 Aug '09

Is it possible to automatically (or non-automatically) install documentation when installing package with cabal-install? Regards

2 2

Library function for map+append
by Dusan Kolar 18 Aug '09

18 Aug '09

Hello all, During a small project I'm trying to develop a small application. It becomes quite often that I need a function mapapp: mapapp _ [] ap = ap mapapp f (a:as) ap = f a : map f as ap I tried hoogle to find such a function with no success. Is there any function/functions built-in "standard" libraries that could easily satisfy the functionality with the same or even better (?) efficiency? Of course, (map f list) ++ append would do the same as mapapp f list append but with less efficiency. Or am I wrong? Thanks Dusan

9 14

Looking up functions inQ monad (Template Haskell).
by Zefirov Sergey 18 Aug '09

18 Aug '09

Why isn't possible to lookup and call user-defined functions while performing Template Haskell actions? Just like the way Template Haskell looks up and calls 'f' in $(f ...). 'f' gets looked up and called. And user of Q monad cannot do that. A colleague of mine, a Lisp user, says that almost everything is in place. It's just not enabled. Is there any problem we do not understand?

2 1

ANN: OpenCLRaw 1.0.1000
by Jeff Heard 18 Aug '09

18 Aug '09

All, I've released a raw binding to the OpenCL platform on Hackage. The main differences between it and the C bindings are that constants have been replaced by newtypes for type safety reasons, void-essential functions that return a token errorcode return a Maybe ErrorCode, and functions that return a handle/ptr and return an error-code through an out-parameter return instead an Either ErrorCode a. Modules are grouped roughly by the section in which they appear in the standard. I'm working on an OpenCL binding that is a little more "cooked" as well as a high-level OpenCL binding. OpenCL is a platform for single-host heterogenous, data-parallel computing that follows roughly the OpenGL and OpenAL conventions for how the library is architected. It will be available by default in the next version of Apple's OS-X, and there are drivers for AMD Opteron/Athlons and nVidia CUDA-based graphics cards. The basic procedure behind running an OpenCL program is: 1. get a list of platforms 2. choose a device on the platform that fits your needs 3. create a context for the platform and the device 4. compile a program and "kernels" written in OpenCL/C 5. create memory buffer objects. 6. execute kernels on the memory buffer objects. 7. rinse hands, repeat. Obviously this is not very functional, but we can't get there without a raw binding. Data Parallel computing is somewhere Haskell is excelling; hopefully this will generate some interest in creating a DPH binding to OpenCL's C99-based language, OpenCL/C.

2 2

Re: [Haskell-cafe] Can't install Haskell Platform on 64-bit Linux
by Colin Paul Adams 18 Aug '09

18 Aug '09

>>>>> "Magnus" == Magnus Therning <magnus(a)therning.org> writes: Magnus> Ouch, there only seems to be a version of 6.10.3 in Fedora Magnus> 11, if I read this correctly: Magnus> https://admin.fedoraproject.org/pkgdb/packages/name/ghc Magnus> (I'm not a Fedora user so I might be completely confused Magnus> here :-) Magnus> It looks like the pre-built ghc can't find the installed Magnus> libgmp. I don't think it's a problem with libgmp. I think that's just a cautionary message, based on experience that this is a common error. -- Colin Adams Preston Lancashire

1 0

Grouping and SIMD in parallel Haskell (using Nested Data Parallel Haskell ideas in legacy code)
by Zefirov Sergey 18 Aug '09

18 Aug '09

I haven't had enough time to polish a paper about the subject, so I decided to post my results here, in Haskell Café. When Simon Peyton-Jones was in Moscow about a month ago I made a bold statement that Parallel Haskell programs, expressed with the help of par and pseq, can be transformed into Nested Data Parallel Haskell. I wrote a simple model of what will be if we group arguments of par and pseq into queues based on parallel arrays from NDPH. We could then evaluate all thunks from that queues in parallel using SIMD (or just getting higher ILP). We also do not have to lock out main spark queue, we lock only queue we should put argument in. The latter turned out to be beneficial in itself, at least in my simple model. Let us look at famous parallel fib function: Fib n | N <= 1 = 1 | otherwise = a `par` b `pseq` a+b where a = fib (n-1) b = fib (n-2) When we evaluate fib we put a spark of a into spark queue and proceed to evaluate b and, then, a+b. When we put a into spark queue, we lock queue out, modify and release. There is a big probability that threads on different cores will compete for queue lock and some of them (most of them) will waste time waiting for lock release. If we group a's into different queue and put to main spark queue a spark to evaluate a complete group of a's at once, we will get less wasted time. This will work even for single CPU. Below is a run of (fib 15) on my model with cpuCount 1, 4 and 16 and a's or b's group length 0 (no grouping) and 16: cpuCount groupLength=0 groupLength=16 modelTicks modelTicks 1 34535 27189 4 12178 7472 16 7568 3157 speedup 2.19 times 3.11 times ticks1/ticks16 I think results speak for itself. I think the idea of `par` argument grouping could be viable. I should note that I made several digressions when I wrote model. One of digressions is that all evaluations are put into queue. In (a `par` b `pseq` a+b) a put into a_queue, b put into b_queue and (a+b) put into main queue. Each evaluated spark update it's "parent" - a spark that wait for it. Also, main loop of single CPU changed from simple (reading main spark queue + execute when get something) into a series of attepmts with fall back on failure: - first read main queue and execute spark if succeed, - else read current a_queue and execute all sparks there if succeed, - else read current b_queue and execute all sparks there if succeed, - else go to main loop. The new (transformed) code for our fib below: -- |Create a new queue based on parallel array. It holds a parallel array with current arguments and a function that performs -- computation in RTS monad (evaluation function). newQueueParArray :: (x -> RTS ()) -> RTSRef ([: x :],x -> RTS ()) a_queue = unsafePerformIO $ newQueueParArray (\x -> fib (x-1)) -- RTSRef (Int,Int -> Int) b_queue = unsafePerformIO $ newQueueParArray (\x -> fib (x-2)) -- RTSRef (Int,Int -> Int) fib n caller | n <= 1 = 1 | otherwise = unsafePerformIO $ do Ab <- addToMainQueue (defer (+) caller) A' <- addToParArrQueue a_queue x ab B' <- addToParArrQueue b_queue x ab addToMainQueue ab -- add a spark to check a' and b' evaluation status, compute a+b and update the caller. That transfomation cannot be done at the source level using usual type (class/families) hackery. It could be done, though, using core-to-core transformations. It is clear that several values of same type (Int for fib) and a function to perform operations over them leads to SIMD execution. I made some provisions to exploit SIMD and ILP in my model. It can load more than single task information and add several values per cycle. This also speeds execution up, but not so radically (about 3%). The source code for model is here: http://82.146.47.211/attachment/wiki/ndpph/ndp-ph.hs (it's a MskHUG.ru domain, we have temporary problems with DNS). It short of comments, but I tried to make understandable function names. Compile it with 'ghc -o ndp-ph --make -O2 ndp-ph' and run with a command line like the following: ndp-ph.exe cpuCount 1 usePrivateGroups 0 maxABTaskLength 64 taskFetchsPerCycle 1 thunksPerAddition 8 cyclesPerAddition 1 I decided to create a model of exexcution instead of modifying an existing implementation because I have not enough time. I wrote it over evenings and a weekend, so it is simple, it's rude and it does the job pretty fine.

2 2

Can't install Haskell Platform on 64-bit Linux
by Colin Paul Adams 18 Aug '09

18 Aug '09

I installed GHC 6.10.4 (as ./configure said it was required) from a .bz2 file on the GHC downloads page. I then tried ./configure for the platform again, and got: checking ghc actually works... no configure: error: Your installation of ghc does not appear to work. It cannot compile a simple program (see config.log for the details). If you installed ghc from a generic binary tarball then it is worth checking that you have the 'gmp' C library and header files installed. (On debian-based systems this package is called libgmp3-dev.) Looking in config.log, I see: configure:2287: checking ghc actually works configure:2296: /usr/local/bin/ghc -o conftest conftest.hs /tmp/ghc22651_0/ghc22651_0.s: Assembler messages: /tmp/ghc22651_0/ghc22651_0.s:43:0: Error: suffix or operands invalid for `push' /tmp/ghc22651_0/ghc22651_0.s:89:0: Error: suffix or operands invalid for `push' /tmp/ghc22651_0/ghc22651_0.s:135:0: Error: suffix or operands invalid for `push' configure:2299: $? = 1 configure: failed program was: main = putStr "Hello world!\n" -- this file generated by TRY-COMPILE-GHC end of failed program. configure:2308: result: no configure:2314: error: Your installation of ghc does not appear to work. It cannot compile a simple program (see config.log for the details). If you installed ghc from a generic binary tarball then it is worth checking that you have the 'gmp' C library and header files installed. (On debian-based systems this package is called libgmp3-dev.) Libgmp is installed. I don't know the significance of the assembler error messages. If I type from the command line: ghci I get the Prelude> prompt, so it looks like the installation is OK at first glance. What should I check next? -- Colin Adams Preston Lancashire

2 2

Request for Comments - hscurrency 0.0.1
by Max Cantor 17 Aug '09

17 Aug '09

Hi all, I'm putting together some simple tools to do safe calculations on different currencies. For instance, making sure that you dont add something like 5 USD + 10 JPY without doing a proper conversion. I've put up some code on google code which probably explains what I'm trying to do more succinctly than this email. I'm curious what poeple think about the library, its the first haskell code I've written for the purpose of sharing and I intend to add it to hackage once I finalize the interface a bit more. The code is at: http://bit.ly/1Cjjlj I'm very open to suggestions on improving the interface. RIght now its very simple and straightforward but potentially limited. I originally wrote it so that the Prc type would be an instance of several numerical classes but eventually stripped that out as it required returning undefined in several cases (such as calling toIntegral on a Prc which was a composite of several currencies). I felt it was better to avoid having partial functions.. Max

5 5

simple hsql question
by Alexander Kotelnikov 17 Aug '09

17 Aug '09

Hi I wanted to see what access to databases HSQL provides and I stumbled in the very beginning. Assume I have a table map1 with attributes "i" and "s" interger and varchar() respectively. The following code fails (with segfault) for me. And I see no other way to tell compiler that I am expecting an interger to be found as 'i' in a fetched row. import Database.HSQL import Database.HSQL.MySQL main :: IO () main = do c <- connect "localhost" "tx_test" "sacha" "" s <- query c "SELECT i FROM map1" print $ getFieldsTypes s i <- (getFieldValue s "i")::IO Int print i disconnect c -- Alexander Kotelnikov Saint-Petersburg, Russia

2 2

Re: Changelogs and "available since"
by Max Desyatov 17 Aug '09

17 Aug '09

Laszlo Nagy <rizsotto(a)gmail.com> writes: > Hi All, > > I was volunteer to solve this problem: > <http://hackage.haskell.org/trac/summer-of-code/ticket/1565> > In a conversation on the libraries(a)haskell.org I was suggested to > scratch my idea here. > > > I would use hoogle for this. Currently it stores the package name and > the symbols of the modules about a package. What do you think about hayoo? I prefer this to hoogle, as hayoo has more complete database across hackage packages, AFAIK > I would extend it to store > the version of the package as well. Extend the query methods to use > all the available package description at search. And make a > VersionRange from the matched Versions. I wrote some code for doing this a while ago. Take a look and make some comments, I can extend this library to satisfy your needs http://hackage.haskell.org/package/cabal-query

3 2