December 2012 - Haskell-Cafe

Fwd: Object Oriented programming for Functional Programmers
by Eli Frey 30 Dec '12

30 Dec '12

sorry, forgot to reply-all ---------- Forwarded message ---------- From: Eli Frey <eli.lee.frey(a)gmail.com> Date: Sun, Dec 30, 2012 at 1:56 PM Subject: Re: [Haskell-cafe] Object Oriented programming for Functional Programmers To: Brandon Allbery <allbery.b(a)gmail.com> > Except not quite... [...] It's not a built in table, it's a hidden parameter. I'll admit this doesn't HAVE to be the implementation. Often time the compiler can monomorphise the types and "perform the lookup" at compile time. But what's going on here tho. > {-# LANGUAGE ExistentialQuantification #-} > data Showable = forall a. Show a => Showable a > > printShowable :: Showable -> IO () > printShowable (Showable x) = print x > > main = mapM printShowable [ Showable "bob", Showable 3, Showable Nothing ] If we take `mapM printShowable` and just give it an arbitrary list, it has to lookup the correct implementations of `print` as it walks down that list at run-time. I believe similar things motivate vtables in c++/java. I don't have a strong intuition about how dynamically typed OO langs deal with this, but I'm sure structural typing has similar issues. On Sun, Dec 30, 2012 at 1:27 PM, Brandon Allbery <allbery.b(a)gmail.com>wrote: > On Sun, Dec 30, 2012 at 3:45 PM, Eli Frey <eli.lee.frey(a)gmail.com> wrote: > >> > mconcat :: Monad m => [m] -> m >> > mconcat = foldl mappend [] >> >> We can think of `mconcat` having a little lookup table inside of itself, >> and whenever we pass it a concrete `[m]`, `mappend` gets looked up and we >> get the implementation for `m`. Typeclasses are just mappings from types >> to functions >> > > Except not quite... the "Monad m =>" in the signature really means "hey, > compiler, pass me the appropriate implementation of Monad so I can figure > out what I'm doing with this type m". It's not a built in table, it's a > hidden parameter. > > Aside from Object Orientation, it is probably a good idea to learn some C >> for a bit too. C is a good language to play in and try and implement more >> advanced language features. Once you reallize that objects are just lookup >> tables of functions bound with a data-structure, you can implement your own >> in C, or you can make closures as functions bundled with (some) of their >> arguments, or you can implement interesting datastructures, or so many >> other fun things. A good understanding of tagged unions has helped me in >> many a convo with an OO head. >> > > A perhaps strange suggestion in this vein: dig up the source code for Xt, > the old X11 Toolkit, and the Xaw widget library that is built atop it. > (It's part of the X11 source tree, since most of the basic X11 utilities > and xterm are based on it.) It implements a primitive object system in C. > Gtk+ does the same, but hides much of the implementation behind macros and > relies on tricky casting etc. behind the scenes for performance; in Xt, the > basic machinery is more easily visible for inspection and much easier to > understand even if you're not all that familiar with C. If you go this > way, once you've figured out what Xt is doing you might go on to see the > more advanced concepts in how Gtk+ does it. > > And once you've done this, you'll have a good idea of what Objective-C and > C++ (minus templates) are doing under the covers. (Mostly C++, since ObjC > is more or less Smalltalk's OO on top of X, whereas the core concepts of > C++ are not so very different from what Xt does.) If you really want to > dig in further, you might want to try to find the source to cfront, the > original C++ implementation which was a preprocessor for the C compiler. > It'll be missing a lot of modern C++ features, but the core is there. > > -- > brandon s allbery kf8nh sine nomine > associates > allbery.b(a)gmail.com > ballbery(a)sinenomine.net > unix, openafs, kerberos, infrastructure, xmonad > http://sinenomine.net >

2 1

ANNOUNCE: bindings to workqueue for distributed applications
by Badi' Abdul-Wahid 30 Dec '12

30 Dec '12

Hello all I'm announcing a release of bindings to the WorkQueue distributed computing framework. This release comprises two packages: - bindings-cctools: low-level FFI bindings http://hackage.haskell.org/package/bindings-cctools - cctools-workqueue: high-level interface built on top of bindings-cctools http://hackage.haskell.org/package/cctools-workqueue WorkQueue is a Master/Worker framework for building flexible, fault-tolerant, distributed applications. WorkQueue has been used to scale to several thousand nodes, combining resources from different execution environments (HPC, cycle-scavenging, Cloud). Details can be found on the project website: http://www.nd.edu/~ccl/software/workqueue The current bindings are for version 3.6.1 of CCTools. Here are documentation for bindings-cctools and cctools-workqueue: http://nd.edu/~cabdulwa/haskell/doc/bindings-cctools-3.6.1.0.1.0.0/html http://nd.edu/~cabdulwa/haskell/doc/cctools-workqueue-3.6.1.0.1.0.0/html A documented example can be found in the repository. http://bitbucket.org/badi/hs-cctools-workqueue/src/v3.6.1.0.1.0.0/Example.hs If you want to try out these bindings, you will need to install CCTools first. http://www.nd.edu/~ccl/software/download.shtml Any feedback is welcome. Thanks. -- Badi' Abdul-Wahid

1 0

Gloss and relatively expensive computations
by Daniel Díaz Casanueva 30 Dec '12

30 Dec '12

Hello cafe! I'm having some performance issues trying to use gloss to make animations where each "model" step is relatively time expensive. I'm using the IO.Simulate module because my step function needs to do an IO operation which takes around 1 second (it may vary). I will try to give some details. I have the step function: step :: Model -> IO Model step m = do ... p <- slowOperation ... return newModel The rest of the program is small and simple. The drawing function (Model -> IO Picture -- it is pure though) is as quick as simple. However, the program get stuck after the first iteration. In the other hand, I am using the animateIO function which ask me for an Int value to set up "the number of simulation steps to take for each second of real time". But I'm more interested in the animation waiting for the simulation step to end. Does it make sense? I would be really thankful with any pointer here. Thanks in advance, Daniel Díaz.

2 2

Re: [Haskell-cafe] [Haskell-beginners] ghc and android
by Nathan Hüsken 30 Dec '12

30 Dec '12

On 12/30/2012 03:43 PM, Brandon Allbery wrote: > On Sun, Dec 30, 2012 at 8:19 AM, Nathan Hüsken <nathan.huesken(a)posteo.de>wrote: > >> Is it currently possible to build haskell programs for android using the >> android NDK? >> If course, I know, bindings have to be written and android GUIs have to be >> written in Java. But if I would write the bindings for the things I need...? >> > > GUIs can be written without requiring Java/Dalvik, at least on the more > recent versions of Android. It's the basic runtime support that is still a > bit lacking, last I checked. Oh, cool. I did not know that. > > >> Via C/gcc? Its not the fastest, but should work, should it not? >> > > It's nonexistent for ARM; to get a working -fvia-C you have to go back to > ghc versions that don't have any clue about ARM (if you want to scare > yourself, check out via-C's Evil Mangler sometime). And I think the hosted > build isn't working at the moment? > > >> Via llvm? >> > > I believe this is how current ARM development is doing it, rather than try > to code up a native code generation backend. (I also get the impression > that the NCGs are going to be deprecated in favor of -fllvm at some point, > although they want to clean up some lingering optimization issues first; > llvm wins for many things but does somewhat poorly for a few.) > > An additional issue: ghci (and Template Haskell because it uses the > bytecode interpreter of ghci internally) currently(?) requires its own > custom linker instead of being able to use the system linker. Said linker > has no support for ARM. The correct fix for this is to redesign ghci so it > doesn't need its own linker; there has been some work in this direction, > but I don't know how complete it is, and it interacts with other issues > such as building Haskell libraries as shared objects. > With ghci, the consol haskell shell is meant, correct? I do not think I would need ghci on android, for what? I think I can live without template haskell.

2 1

ghc and android
by Nathan Hüsken 30 Dec '12

30 Dec '12

As suggested on the haskell beginners list, I am taking this to haskell-cafe. Hey, I know, this question has been asked in the past and there is a lot information one can find in the net. But in all of this information, I can not find definite answers for this. Is it currently possible to build haskell programs for android using the android NDK? If course, I know, bindings have to be written and android GUIs have to be written in Java. But if I would write the bindings for the things I need...? Via C/gcc? Its not the fastest, but should work, should it not? Via llvm? If not, what is the problem in doing this? Thank you! Nathan _______________________________________________ Beginners mailing list Beginners(a)haskell.org http://www.haskell.org/mailman/listinfo/beginners

1 0

Retiring as maintainer of dataenc
by Magnus Therning 29 Dec '12

29 Dec '12

Hi all, A few years ago I found a need to base64 encode a bit of text inside vim and realised that there was no convenient way of doing it. This resulted in dataenc[1] and a pair of tools built on it, omnicodec[2]. The main goals were to make dataenc correct, support many encodings, and easy to maintain. This resulted in very simple (some would say naïve) code, with fairly good test coverage. As some have noticed, this simplicity in algorithm results in rather bad performance on large data. Which is why I've been planning for a replacement for a while now. This replacement is now here, sandi[3]. The goals of sandi are similar to dataenc, correct, many encodings, fairly simple, but also reasonably fast. This means that I personally think dataenc has reached the end of its life and I won't spend any more time on maintaining it. If anyone is interested in taking over maintenance then please announce it on the list. /M [1]: http://hackage.haskell.org/package/dataenc [2]: http://hackage.haskell.org/package/omnicodec [3]: http://hackage.haskell.org/package/sandi -- Magnus Therning OpenPGP: 0xAB4DFBA4 email: magnus(a)therning.org jabber: magnus(a)therning.org twitter: magthe http://therning.org/magnus I invented the term Object-Oriented, and I can tell you I did not have C++ in mind. -- Alan Kay

1 0

ANN: sandi 0.1.1 - a data encoding/decoding library
by Magnus Therning 29 Dec '12

29 Dec '12

The other day I released version 0.1.1 of sandi, a data encoding/decoding library supporting the following encodings: base16, base32, base32hex, base64, base64url, base85, quoted-printable, uuencode, xxencode, yEnc I believe it's reasonably fast, but I still have to look more closely at criterion to get some hard numbers confirming that :) More details can be found at http://hackage.haskell.org/package/sandi /M -- Magnus Therning OpenPGP: 0xAB4DFBA4 email: magnus(a)therning.org jabber: magnus(a)therning.org twitter: magthe http://therning.org/magnus Perl is another example of filling a tiny, short-term need, and then being a real problem in the longer term. -- Alan Kay

1 0

getting upload date from hackage-db?
by Patrick Pelletier 29 Dec '12

29 Dec '12

(Firstly, I apologize if this is not the right mailing list for this question. But haskell-cafe seems to be the place to go when in doubt.) I'm using the hackage-db package (Distribution.Hackage.DB) to analyze Hackage programatically. It parses 00-index.tar and provides a mapping from package name and version to a GenericPackageDescription. I've figured out how to get most of the information I want from the GenericPackageDescription. But one piece of information that's present on the Hackage web pages that I'd like to be able to get is "Upload date", but I can't find this information anywhere in the GenericPackageDescription. Am I missing it? It looks like the modification dates of the cabal files inside 00- index.tar correspond to the upload date. Is this correct? If so, I could parse this out of 00-index.tar myself (using the "tar" package). Is that the best way to do what I'm trying to accomplish, or is there a better, less "reinventing the wheel" way that I'm missing? Thanks, --Patrick

1 0

ANN: shelltestrunner 1.3
by Simon Michael 28 Dec '12

28 Dec '12

I'm pleased to announce the release of shelltestrunner 1.3. New in this release: * support latest Diff, cmdargs, test-framework; tested with GHC 7.6.1 (Magnus Therning) * fix unicode handling on GHC >= 7.2 shelltestrunner tests command-line programs (or arbitrary shell commands.) It reads simple declarative tests specifying a command, some input, and the expected output, and can run them run in parallel, selectively, with a timeout, in color, and/or with differences highlighted. shelltestrunner has been tested on gnu/linux, mac and windows; projects using it include hledger, berp, cblrepo and eddie. shelltestrunner is free software released under GPLv3+. Home: http://joyful.com/shelltestrunner Install: $ cabal install shelltestrunner

1 0

my Fasta is slow ;(
by Branimir Maksimovic 28 Dec '12

28 Dec '12

This time I have tried fasta benchmark since current entries does notdisplay correct output.Program is copy of mine http://benchmarksgame.alioth.debian.org/u64q/program.php?test=fasta&lang=gp… benchmark, but unfortunately executes more than twice time. Seems to me that culprit is in function random as I have tested rest of codeand didn't found speed related problems. bmaxa@maxa:~/shootout/fasta$ time ./fastahs 25000000 > /dev/null real 0m5.262suser 0m5.228ssys 0m0.020s bmaxa@maxa:~/shootout/fasta$ time ./fastacpp 25000000 > /dev/null real 0m2.075suser 0m2.056ssys 0m0.012s Since I am planning to contribute program, perhaps someone cansee a problem to speed it up at least around 3.5 secs which is speed of bench that display incorrect result (in 7.6.1). Program follows: {-# LANGUAGE BangPatterns #-}{- The Computer Language Benchmarks Game http://shootout.alioth.debian.org/ contributed by Branimir Maksimovic-} import System.Environmentimport System.IO.Unsafe import Data.IORefimport Data.Array.Unboxedimport Data.Array.Storableimport Data.Array.Baseimport Data.Word import Foreign.Ptrimport Foreign.C.Types type A = UArray Int Word8type B = StorableArray Int Word8type C = (UArray Int Word8,UArray Int Double) foreign import ccall unsafe "stdio.h" puts :: Ptr a -> IO ()foreign import ccall unsafe "string.h" strlen :: Ptr a -> IO CInt main :: IO () main = do n <- getArgs >>= readIO.head let !a = (listArray (0,(length alu)-1) $ map (fromIntegral. fromEnum) alu:: A) make "ONE" "Homo sapiens alu" (n*2) $ Main.repeat a (length alu) make "TWO" "IUB ambiguity codes" (n*3) $ random iub make "THREE" "Homo sapiens frequency" (n*5) $ random homosapiens make :: String -> String -> Int -> IO Word8 -> IO (){-# INLINE make #-}make id desc n f = do let lst = ">" ++ id ++ " " ++ desc a <- (newListArray (0,length lst) $ map (fromIntegral. fromEnum) lst:: IO B) unsafeWrite a (length lst) 0 pr a make' n 0 where make' :: Int -> Int -> IO () make' !n !i = do let line = (unsafePerformIO $ newArray (0,60) 0 :: B) if n > 0 then do !c <- f unsafeWrite line i c if i+1 >= 60 then do pr line make' (n-1) 0 else make' (n-1) (i+1) else do unsafeWrite line i 0 l <- len line if l /= 0 then pr line else return () pr :: B -> IO ()pr line = withStorableArray line (\ptr -> puts ptr)len :: B -> IO CIntlen line = withStorableArray line (\ptr -> strlen ptr) repeat :: A -> Int -> IO Word8repeat xs !n = do let v = unsafePerformIO $ newIORef 0 !i <- readIORef v if i+1 >= n then writeIORef v 0 else writeIORef v (i+1) return $ xs `unsafeAt` i random :: C -> IO Word8random (a,b) = do !rnd <- rand let find :: Int -> IO Word8 find !i = let !c = a `unsafeAt` i !p = b `unsafeAt` i in if p >= rnd then return c else find (i+1) find 0 rand :: IO Double{-# INLINE rand #-}rand = do !seed <- readIORef last let newseed = (seed * ia + ic) `rem` im newran = fromIntegral newseed * rimd rimd = 1.0 / (fromIntegral im) im, ia, ic :: Int im = 139968 ia = 3877 ic = 29573 writeIORef last newseed return newran where last = unsafePerformIO $ newIORef 42 alu :: [Char] alu = "GGCCGGGCGCGGTGGCTCACGCCTGTAATCCCAGCACTTTGG\ \GAGGCCGAGGCGGGCGGATCACCTGAGGTCAGGAGTTCGAGA\ \CCAGCCTGGCCAACATGGTGAAACCCCGTCTCTACTAAAAAT\ \ACAAAAATTAGCCGGGCGTGGTGGCGCGCGCCTGTAATCCCA\ \GCTACTCGGGAGGCTGAGGCAGGAGAATCGCTTGAACCCGGG\ \AGGCGGAGGTTGCAGTGAGCCGAGATCGCGCCACTGCACTCC\ \AGCCTGGGCGACAGAGCGAGACTCCGTCTCAAAAA" mkCum :: [(Char,Double)] -> [(Word8,Double)]mkCum lst = map (\(c,p) -> ((fromIntegral.fromEnum) c,p)) $ scanl1 (\(_,p) (c',p') -> (c', p+p')) lst homosapiens, iub :: C iub' = mkCum [('a',0.27),('c',0.12),('g',0.12),('t',0.27),('B',0.02) ,('D',0.02),('H',0.02),('K',0.02),('M',0.02),('N',0.02) ,('R',0.02),('S',0.02),('V',0.02),('W',0.02),('Y',0.02)] homosapiens' = mkCum [('a',0.3029549426680),('c',0.1979883004921) ,('g',0.1975473066391),('t',0.3015094502008)] iub = (listArray (0, (length iub')-1) $ map fst iub', listArray (0, (length iub')-1) $ map snd iub') homosapiens = (listArray (0, (length homosapiens')-1) $ map fst homosapiens', listArray (0, (length homosapiens')-1) $ map snd homosapiens')

2 4