- Haskell-Cafe - Haskell.org

Re: [Haskell-cafe] Atom - help.
by Tom Hawkins 28 Nov '09

28 Nov '09

Calling foreign functions that return values is problematic for the compiler -- I haven't invested enough time to come up with a good solution. The work around is to assign the result to an external variable. The drawback is the result will not be available until the rule executing the action has completed. results <- bool' "result" -- Global, external variable to capture result. action (\ [a, b, c] -> printf "result = function(%s, %s %s)" a b c) [a, b, c] -Tom On Sat, Jun 13, 2009 at 5:03 AM, Radamés Ajna<radamajna(a)gmail.com> wrote: > hi there, > > I've been working with haskell and atom dsl, however I'm new to haskell.. > John Van Enk inspired me with programming arduino->atmegas , with haskell, > So here is my problem. > With an action , action :: ([String] -> String) -> [UE] -> Atom () , > it's possible make calls to some precoded C functions like, > > setLED v = action (\[x] -> "setLED(" ++ x ++ ")") [v'] > where v' = ue . value $ v > > where setLed is a precode function. > > I want call a precoded C function that returns some value, but action > always give me Atom () , > that's I would like to do but I don't know how .. > > Atom (V Int8) < ---- "int read()" > > > thanks... > related links > http://hackage.haskell.org/package/atom-0.0.5 > http://blog.sw17ch.com/wordpress/?p=111 > http://leepike.wordpress.com/2009/05/05/an-atomic-fibonacci-server-explorin… > > > > Radamés > _______________________________________________ > Haskell-Cafe mailing list > Haskell-Cafe(a)haskell.org > http://www.haskell.org/mailman/listinfo/haskell-cafe >

2 2

ANN: package-vt-0.1.3.3, Haskell Package Versioning Tool
by Krzysztof Skrzętnicki 28 Nov '09

28 Nov '09

Hi! I'm pleased to announce the first release of my program package-vt. It's role is to detect changes between two versions of package that should cause version bump. Recommended versioning policy for Haskell packages is on the wiki [1]. The ultimate goal is to get as good approximation of the algorithm specified there as possible. Currently the tool is far from being complete: it only detects changes to exports in .cabal and module files, but doesn't take any types into consideration, not to mention other possible heuristics. This tool is not meant as a replacement for developer's own brain. It can only detect a subset of changes, and there will always be some important changes it won't catch, like adding such line: head (_:y:_) = y However, it can be used for some quality checks. Thing that needs improving is user interface. No options are currently accepted. This, and a lot more, will change in future versions. You can find a Git repo here [2]. The repo contains 'tests' directory with two useful scripts. You can use them to run package-vt on anything from Hackage. Here are a few random usage examples: 1) Detect small change within module package-vt packages/HCL/1.1/HCL-1.1/HCL.cabal packages/HCL/1.2/HCL-1.2/HCL.cabal Single changes: ChangeABC : Entity added to module ModuleName "HCL" --- EVar (UnQual (Ident "prompt1")) --- Biggest change comes from: ChangeABC : Entity added to module ModuleName "HCL" --- EVar (UnQual (Ident "prompt1")) 2) Detect new module added to .cabal package-vt 0.2.0/AERN-Net-0.2.0/AERN-Net.cabal 0.2.1/AERN-Net-0.2.1/AERN-Net.cabal Single changes: ChangeABC : Entity added to module ModuleName "Control.ERNet.Foundations.Process" --- EThingAll (UnQual (Ident "ERProcessAction")) ---ChangeABC : Module(s) added.(Consider AB change if new modules are likely to cause name collisions) --- "Control/ERNet/Blocks/Real/LFT" --- Biggest change comes from: ChangeABC : Module(s) added.(Consider AB change if new modules are likely to cause name collisions) --- "Control/ERNet/Blocks/Real/LFT" --- 3) Using testing script: package-vt correctly detects that AERN-RnToRM-0.4.9 should rather be 0.5.0 because the module Data.Number.ER.RnToRm.UnitDom.ChebyshevBase.Polynom.Field has been removed. ./test-on-package.lhs AERN-RnToRm --cut-- >>> Testing package-vt on AERN-RnToRm-0.4.2 and AERN-RnToRm-0.4.9 Parse error: VarSym "#" : SrcLoc {srcFilename = "0.4.2/AERN-RnToRm-0.4.2/src/Data/Number/ER/RnToRm/Approx/DomTransl.hs", srcLine = 174, srcColumn = 1} Single changes: ChangeABC : Entity added to module ModuleName "Data.Number.ER.RnToRm.Approx" --- EVar (UnQual (Ident "keyPointsConsistencyCheck")) EVar (UnQual (Ident "keyPointsPointwiseConsistencyCheck")) ---ChangeABC : Entity added to module ModuleName "Data.Number.ER.RnToRm.UnitDom.Approx" --- EVar (UnQual (Ident "keyPointsConsistencyCheck")) EVar (UnQual (Ident "keyPointsPointwiseConsistencyCheck")) ---ChangeAB : Module(s) removed/renamed. --- "Data/Number/ER/RnToRm/UnitDom/ChebyshevBase/Polynom/Field" --- Biggest change comes from: ChangeAB : Module(s) removed/renamed. --- "Data/Number/ER/RnToRm/UnitDom/ChebyshevBase/Polynom/Field" --- ExitSuccess --cut-- In case of any trouble please email me or report an issue here [3]. In the near future I'll post more specific plans about development on my blog [4]. I hope you will find it useful. Best regards, Krzysztof Skrzętnicki [1] http://www.haskell.org/haskellwiki/Package_versioning_policy [2] http://github.com/Tener/haskell-package-vt [3] http://github.com/Tener/haskell-package-vt/issues [4] http://mostlycode.wordpress.com

3 2

GLFW Textures (bloxorz)
by jean legrand 27 Nov '09

27 Nov '09

Hi Haskellers, I have two issues with the bloxorz game (current version bloxorz-0.1 on hackage). Perhaps they're mutually related. You tell me. First, it seems that there is no refresh on the successive block positions. Look at http://img524.imageshack.us/img524/5945/bloxissue1.png to see what I mean. Second, any resize gives me a black window. Everything else is fine (levels, game rules, keys ...) To produce the exe, I did : ghc --make -O2 -o bloxorz Main.hs and, as I think it deals with the textures, here is what I get with ghci : Loading package syb ... linking ... done. Loading package base-3.0.3.1 ... linking ... done. Loading package ObjectName-1.0.0.0 ... linking ... done. Loading package OpenGLRaw-1.1.0.1 ... linking ... done. Loading package GLURaw-1.1.0.0 ... linking ... done. Loading package StateVar-1.0.0.0 ... linking ... done. Loading package Tensor-1.0.0.1 ... linking ... done. Loading package OpenGL-2.4.0.1 ... linking ... done. Loading package GLFW-0.4.1 ... linking ... done. and I use GHC6.10.4 with Kubuntu 9.10. Any ideas ?

2 2

Re: [Haskell] ANN: NoSlow - Microbenchmarks for array libraries
by Henning Thielemann 27 Nov '09

27 Nov '09

On Fri, 27 Nov 2009, Roman Leshchinskiy wrote: > It compiles and runs (using Brian's criterion) a fairly random collection of very small loop kernels for each of those and produces a lot of data which it then uses to generate ugly HTML tables. In the future, it will have more benchmarks, more complex benchmarks and much prettier tables. > > You can get more information (including the ugly tables) from my blog > > http://unlines.wordpress.com/2009/11/27/noslow > > and NoSlow itself from Hackage > > http://hackage.haskell.org/package/NoSlow Is there also a darcs repository? I want to note, that storablevector requires Quickcheck only for tests. In the darcs repository, this is better handled by the Cabal file: http://code.haskell.org/storablevector/ There you get the dependency on QuickCheck (1) only when building with buildTests flag set. Now I have to get NoSlow running, but surprisingly it needs cairo and gtk, which I assume are not so easy to install.

2 1

Other instances of Integral
by Tsunkiet Man 27 Nov '09

27 Nov '09

Hello, I would like to ask wheter there are other instances of the class Integral? And I would like to ask what the difference is between the following functions: SomeFunctionA :: (Integral a) => a -> a, SomeFunctionB (Integer a) => a -> a, SomeFunctionC (Int a) => a -> a. What I do know is, that the Int can have underflow and overflows, however I don't actually see the difference (and I can't really find a difference on Google as it gives me results that aren't really relevant to my question) between prefering to use an Integral a when I've already got an Integer. (Assuming I didn't missed the definiton of a Integral, which has by definition (I looked it up on Google: http://www.zvon.org/other/haskell/Outputprelude/Integral_c.html) has two instances) What I also know is that operatons on a Int is faster than operations on a Integral or Integer. Can someone explain to me what kind of advantages and disadvantages I would get when substituting SomeFunctionB for someFunctionA? Thank you for your help!

5 5

Reducing Latency - Reducing the Size of Larger Data Structures
by Casey Hawthorne 27 Nov '09

27 Nov '09

An idea to reduce memeory latency, is to have a complex dual/quad/etc. core CPU surrounded by four or more simpler CPUs. The simpler CPUs have a simpler instruction set and do less context switching. So any operations (e.g. folding, filtering) that reduce the size of a larger data structure and that can be done on the simpler CPU instruction sets are shunted off to the simpler CPUs. So, the complex CPUs receive less data and therefore can have better cacheing performance, less latency and better throughput. -- Regards, Casey

1 0

Re: [Haskell] ANN: NoSlow - Microbenchmarks for array libraries
by Henning Thielemann 27 Nov '09

27 Nov '09

Roman Leshchinskiy schrieb: > I'm pleased to announce NoSlow, a nascent benchmark suite for various array (and list) libraries with particular emphasis on finding out how well they fuse things. At the moment, it knows about > > * standard lists > * primitive DPH arrays (dph-prim-seq) > * uvector > * vector > * storablevector > > It compiles and runs (using Brian's criterion) a fairly random collection of very small loop kernels for each of those and produces a lot of data which it then uses to generate ugly HTML tables. In the future, it will have more benchmarks, more complex benchmarks and much prettier tables. > > You can get more information (including the ugly tables) from my blog > > http://unlines.wordpress.com/2009/11/27/noslow > > and NoSlow itself from Hackage > > http://hackage.haskell.org/package/NoSlow > > Any help would be highly appreciated as I don't have a lot of time to work on it. Since storablevector is in no way optimized for fusion, I'd like to see how well it compares with Fusion-Streams that are fused into strict and lazy storablevectors by http://hackage.haskell.org/package/storablevector-streamfusion . Thanks for the benchmarking effort!

1 0

Re: [Haskell] ANN: NoSlow - Microbenchmarks for array libraries
by Henning Thielemann 27 Nov '09

27 Nov '09

Roman Leshchinskiy schrieb: > I'm pleased to announce NoSlow, a nascent benchmark suite for various array (and list) libraries with particular emphasis on finding out how well they fuse things. At the moment, it knows about > > * standard lists > * primitive DPH arrays (dph-prim-seq) > * uvector > * vector > * storablevector > > It compiles and runs (using Brian's criterion) a fairly random collection of very small loop kernels for each of those and produces a lot of data which it then uses to generate ugly HTML tables. In the future, it will have more benchmarks, more complex benchmarks and much prettier tables. > > You can get more information (including the ugly tables) from my blog > > http://unlines.wordpress.com/2009/11/27/noslow > > and NoSlow itself from Hackage > > http://hackage.haskell.org/package/NoSlow > > Any help would be highly appreciated as I don't have a lot of time to work on it. Since storablevector is in no way optimized for fusion, I'd like to see how well it compares with Fusion-Streams that are fused into strict and lazy storablevectors by http://hackage.haskell.org/package/storablevector-streamfusion . Thanks for the benchmarking effort!

1 0

Show function a -> String
by Tsunkiet Man 27 Nov '09

27 Nov '09

Hello, let's say I'm defining function that does the following: ShowFloat -> Floating -> String, at what kind of userinput (assuming you insert some floating number) will I get an error? (Please do not post overflow, because I already know that =)). Thanks for your help again guys!

4 5

ANN: wumpus-core-0.13.0
by Stephen Tetley 27 Nov '09

27 Nov '09

Hello All Wumpus-core is a library for drawing 2D vector pictures, supporting output to PostScript and SVG. It has no FFI dependencies, so should be portable across operating systems (assuming GHC with type families). Wumpus-core is functional, though quite low-level - its similar to PostScript but with a more 'coordinate free' style. I'm working on a complementary package for higher-level stuff, but it will be some time before it is release ready. Many thanks to Conal Elliott and Iavor S. Diatchki for their invaluable libraries (VectorSpace and MonadLib/xml respectively). Best wishes Stephen http://hackage.haskell.org/package/wumpus-core

2 2