- Haskell-Cafe - Haskell.org

[Long, probably not-beginners anymore] Parallel folds and folds as arrows (was: Re: [Haskell-beginners] Re: When, if ever, does Haskell "calculate once"?)
by Maciej Piechotka 07 May '10

07 May '10

On Thu, 2010-05-06 at 23:46 +0200, Daniel Fischer wrote: > Share.share :: GHC.Types.Int > GblId > [Str: DmdType] > Share.share = > case GHC.List.$wlen @ GHC.Integer.Type.Integer Share.share_a 0 > of ww_amc { __DEFAULT -> > GHC.Types.I# (GHC.Prim.+# ww_amc ww_amc) > } > Hmm. What's the name of this form and how to get it? > > No. > > cFoldl' f g (b0,c0) xs0 = lgo b0 c0 xs0 > where > lgo b c [] = (b,c) > lgo !b !c (x:xs) = lgo (f b x) (g c x) xs > Ok. Fixed (I tried fast rewrite from foldr') > > 64-bit system? I get > 64 bit, GHC 6.12.2. % ghc -V The Glorious Glasgow Haskell Compilation System, version 6.12.2 % file a.out a.out: ELF 64-bit LSB executable, x86-64, version 1 (SYSV), dynamically linked (uses shared libs), for GNU/Linux 2.6.9, not stripped > > And that is strange, because I get the same figures for that one as for the > first (times differ by a few hundredths of a second). Fixed or non-fixed version? > Is that a difference between 32-bit code generator and 64-bit or between > GHC versions (6.12.2 here, but 6.10.3 gives roughly the same results)? > Hmm. Compiler and platform matches. Unless you use some other 64-bit platform - not x86-64 ;) > > > > main = > > > print $! uncurry (+) (cFoldl' lengthFold lengthFold > > > (0, 0) [1..size]) > > And that gives the same figures as the other two (plus/minus 0.05s). > > > > > All are compiled with optimizations. > > All compiled with -O2. > Hmm. Difference between -O1 and -O2 Fixed versions and with sum (and -O2): > main = let a = [1..size] > l = length a + sum a > in print $! l Lot's of memory(Over 3 GiB). I voluntarily killed process > main = print $! length [1..size] + sum [1..size] Lot's of memory(Over 3 GiB). I voluntarily killed process. So far as being inplace. > main = print $! uncurry (+) (cFoldl' lengthFold (+) (0, 0) [1..size]) 5000000150000000 16,889,753,976 bytes allocated in the heap 3,356,480 bytes copied during GC 1,976 bytes maximum residency (1 sample(s)) 28,200 bytes maximum slop 1 MB total memory in use (0 MB lost due to fragmentation) Generation 0: 32216 collections, 0 parallel, 0.29s, 0.38s elapsed Generation 1: 1 collections, 0 parallel, 0.00s, 0.00s elapsed INIT time 0.00s ( 0.00s elapsed) MUT time 14.89s ( 15.12s elapsed) GC time 0.29s ( 0.38s elapsed) EXIT time 0.00s ( 0.00s elapsed) Total time 15.18s ( 15.50s elapsed) %GC time 1.9% (2.4% elapsed) Alloc rate 1,134,094,110 bytes per MUT second Productivity 98.1% of total user, 96.1% of total elapsed ./a.out +RTS -s 15.18s user 0.11s system 98% cpu 15.503 total Lowered to size = 100000 (bigger causes stack overflow in first main): > main = let a = [1..size] > l = length a + sum a > in print $! l 5000150000 22,045,352 bytes allocated in the heap 18,781,768 bytes copied during GC 6,316,904 bytes maximum residency (4 sample(s)) 3,141,912 bytes maximum slop 17 MB total memory in use (0 MB lost due to fragmentation) Generation 0: 23 collections, 0 parallel, 0.05s, 0.05s elapsed Generation 1: 4 collections, 0 parallel, 0.03s, 0.04s elapsed INIT time 0.00s ( 0.00s elapsed) MUT time 0.02s ( 0.03s elapsed) GC time 0.08s ( 0.09s elapsed) EXIT time 0.00s ( 0.00s elapsed) Total time 0.10s ( 0.12s elapsed) %GC time 78.4% (75.8% elapsed) Alloc rate 1,002,334,818 bytes per MUT second Productivity 20.6% of total user, 17.2% of total elapsed ./a.out +RTS -s 0.10s user 0.02s system 96% cpu 0.129 total > main = print $! length [1..size] + sum [1..size] 5000150000 30,077,024 bytes allocated in the heap 17,482,888 bytes copied during GC 5,602,600 bytes maximum residency (4 sample(s)) 3,144,312 bytes maximum slop 15 MB total memory in use (0 MB lost due to fragmentation) Generation 0: 38 collections, 0 parallel, 0.04s, 0.05s elapsed Generation 1: 4 collections, 0 parallel, 0.03s, 0.03s elapsed INIT time 0.00s ( 0.00s elapsed) MUT time 0.02s ( 0.03s elapsed) GC time 0.07s ( 0.08s elapsed) EXIT time 0.00s ( 0.00s elapsed) Total time 0.09s ( 0.11s elapsed) %GC time 72.0% (70.0% elapsed) Alloc rate 1,156,986,613 bytes per MUT second Productivity 26.9% of total user, 22.9% of total elapsed ./a.out +RTS -s 0.09s user 0.02s system 97% cpu 0.116 total > main = print $! uncurry (+) (cFoldl' lengthFold (+) (0, 0) [1..size]) 5000150000 17,128,128 bytes allocated in the heap 10,608 bytes copied during GC 2,072 bytes maximum residency (1 sample(s)) 28,024 bytes maximum slop 1 MB total memory in use (0 MB lost due to fragmentation) Generation 0: 32 collections, 0 parallel, 0.00s, 0.00s elapsed Generation 1: 1 collections, 0 parallel, 0.00s, 0.00s elapsed INIT time 0.00s ( 0.00s elapsed) MUT time 0.02s ( 0.02s elapsed) GC time 0.00s ( 0.00s elapsed) EXIT time 0.00s ( 0.00s elapsed) Total time 0.02s ( 0.02s elapsed) %GC time 5.3% (3.9% elapsed) Alloc rate 951,721,286 bytes per MUT second Productivity 94.7% of total user, 90.2% of total elapsed ./a.out +RTS -s 0.02s user 0.00s system 96% cpu 0.024 total It seems that the best, at least for large inputs. ----------------------------------------------------------------------- On the other hand it seems to form an arrow[1]. First the result of test: 5000000150000000 12,800,063,440 bytes allocated in the heap 2,545,048 bytes copied during GC 1,968 bytes maximum residency (1 sample(s)) 28,216 bytes maximum slop 1 MB total memory in use (0 MB lost due to fragmentation) Generation 0: 24414 collections, 0 parallel, 0.24s, 0.29s elapsed Generation 1: 1 collections, 0 parallel, 0.00s, 0.00s elapsed INIT time 0.00s ( 0.00s elapsed) MUT time 7.18s ( 7.35s elapsed) GC time 0.24s ( 0.29s elapsed) EXIT time 0.00s ( 0.00s elapsed) Total time 7.42s ( 7.64s elapsed) %GC time 3.3% (3.8% elapsed) Alloc rate 1,782,759,994 bytes per MUT second Productivity 96.7% of total user, 93.9% of total elapsed ./a.out +RTS -s 7.42s user 0.09s system 98% cpu 7.648 total (Yes - lower then the code above - result reproducable). Code: > {-# LANGUAGE BangPatterns #-} > import Control.Arrow > import Control.Category > import Data.List > import Prelude hiding (id, (.)) > > newtype FoldlArrow a b c = FoldlArrow (a -> b -> c) > > instance Category (FoldlArrow a) where > id = FoldlArrow $ \a !b -> b > (FoldlArrow f) . (FoldlArrow g) = > FoldlArrow $ \a !b -> let !c = g a b > in f a c > > instance Arrow (FoldlArrow a) where > arr f = FoldlArrow $ const f > first (FoldlArrow f) = FoldlArrow $ \a (!b, !c) -> (f a b, c) > second (FoldlArrow f) = FoldlArrow $ \a (!b, !c) -> (b, f a c) > (FoldlArrow f) *** (FoldlArrow g) = > FoldlArrow $ \a (!b, !c) -> (f a b, g a c) > (FoldlArrow f) &&& (FoldlArrow g) = > FoldlArrow $ \a !b -> (f a b, g a b) > > instance ArrowChoice (FoldlArrow a) where > left (FoldlArrow f) = FoldlArrow left' > where left' a (Left !l) = (Left $! f a l) > left' a (Right !r) = (Right r) > right (FoldlArrow f) = FoldlArrow right' > where right' a (Left !l) = (Left l) > right' a (Right !r) = > (Right $! f a r) > (FoldlArrow f) +++ (FoldlArrow g) = > FoldlArrow choice > where choice a (Left !l) = (Left $! f a l) > choice a (Right !r) = (Right $! g a r) > (FoldlArrow f) ||| (FoldlArrow g) = > FoldlArrow choice > where choice a (Left !l) = f a l > choice a (Right !r) = g a r > > instance ArrowApply (FoldlArrow a) where > app = FoldlArrow $ \a (FoldlArrow !f, !b) -> f a b > > doFoldl :: FoldlArrow a b b -> b -> [a] -> b > doFoldl (FoldlArrow f) = foldl' (flip f) > > element :: FoldlArrow a b a > element = FoldlArrow const > > lengthA :: FoldlArrow a Int Int > lengthA = arr (+1) > > sumA :: Num a => FoldlArrow a a a > sumA = arr (uncurry (+)) . (element &&& id) > > size = 100000000 > > main = > print $! uncurry (+) $ doFoldl (lengthA *** sumA) (0, 0) [1..size] On the other hand if I use arrow syntax: > myArr :: Num a => FoldlArrow a (a, a) (a, a) > myArr = proc (l, s) -> do e <- element -< () > returnA -< (l + 1, s + e) It starts consuming memory as well. Somewhere is lazy passing of value but I cannot find where Regards PS. As it is probably out of scope and topic of beginners mailing list I'm CC'ing cafe (possibly beginners should be dropped). [1] It can be extended to work on other arrows as well - not only (->).

2 3

ANNOUNCE: gt-tools-0.1.4
by Sergei Trofimovich 07 May '10

07 May '10

Hey, -cafe! I'm glad to announce release of Google Translate Service[1] Tools[2]. Currently it's a command line tool 'gtc' (queen of this post) and 'gtg' (more a hello-world type gtk2hs user). So let's play with that beast a little! PREREQUISITES * ghs-6.12+ - gt-tools relies on proper locale detection for passed command line arguments INSTALLATION $ cabal update Downloading the latest package list from hackage.haskell.org $ cabal install gt-tools -f-gui # avoid 'gtg' for now ... Done! It was simple. USE $ gtc # sniff the beast Usage: Standart mode: gtc 'fromLang' 'toLang' 'text' Interactive mode: gtc -i 'fromLang' 'toLang' Where 'Lang' is 2 sumbols code, i.e. be, en, ru. Commandline keys: --interactive or -i - interactive mode. --help or -h - to see this help. --list or -l - list of available languages. $ gtc -l # language list # a lot $ gtc en no "good evening" Source language: en Translation: god kveld Original text: good evening Translit: Whoa! $ gtc -i en fi # interactive mode! > # waits for prompt! > who are you? -------------------------------------------------------------------------------- Source language: en Translation: kuka sinä olet? Original text: who are you? Translit: # leading exclamation'!' mark allows you to invert source and destination language > !päivää -------------------------------------------------------------------------------- Source language: fi Translation: days Original text: päivää Translit: interjection: good afternoon Integration with other translating services is considered by author as well as user interface should be polished a little more too. Any feedback is appreciated! /me wonders if Miss lambdabot might like to have such functionality. What do you think? Have fun! [1] It's a simple translation service: http://translate.google.com/ [2] "home page" for gt-tools: http://github.com/styx/gtc -- Sergei

4 3

Database connection pool
by Michael Snoyman 07 May '10

07 May '10

Hi all, I would like to pool my database connections in an application I'm writing, and so far haven't found any prior art on the subject (besides this[1]). I was wondering if: * There's a package somewhere that does this * Others have implemented it and have suggestions * There's some big gotchas The stackoverflow that I linked to gives one approach to this that seems valid. My original thought on an implementation was to use a Chan. That way: * When a connection is released, is goes to the end of the pool, so connections get used evenly (not sure if this actually matters in practice). * If a process dies and for some reason doesn't return its connection, there shouldn't be a problem. Of course, I'd use bracket, so such a situation should never arise anyway. * Keep the coding against MVar to a minimum to avoid normal multi-threaded pitfalls. I'm using HDBC in this case, but I don't think I'll be using any HDBC-specific code. Thanks, Michael [1] http://stackoverflow.com/questions/1141677/concurrent-db-connection-pool-in…

3 7

Mirror repository for gtk2hs
by Andy Stewart 07 May '10

07 May '10

Hi all, Because code.haskell.org is unstable recently, and many gtk2hs users can't get code from http://code.haskell.org/gtk2hs So i build a mirror repository at http://patch-tag.com/r/AndyStewart/gtk2hs-sync-mirror/home You can access this mirror repository when code.haskell.org down. This mirror repository synchronous http://code.haskell.org/gtk2hs Hourly. Because mirror repository just read-only, please send any patches to gtk2hs main repository (http://code.haskell.org/gtk2hs) I need thanks "Thomas Hartman" (the administor of patch-tag.com) i can't build this mirror repository if haven't his help. Thanks Thomas! Enjoy, everybody! -- Andy

1 0

"Dynamic" CSV Parsing - Parsec
by Günther Schmidt 07 May '10

07 May '10

Hello, I'm trying to build a CSV parser that can dynamically assemble a parser from the values of the first line. As the most simple example the parse of the first line would return a parser with which subsequent lines would then be parsed. This parser would, for instance, only parse lines with the exact number of columns as found in the first line. Where I eventually want to go is a bit more complicated than this, but for now I'd be grateful for suggestions on how to go about the simple case. Best regards Günther

2 1

Fwd: Error instaling Happstack on Windows - cabal bug?
by Flavio Botelho 06 May '10

06 May '10

Trying to install on Windows, i am having a strange problem (may actually be some cabal bug?), More specifically trying to install happstack-util [13 of 19] Compiling Happstack.Crypto.MD5 ( src\Happstack\Crypto\MD5.hs, dist\bu ild\Happstack\Crypto\MD5.o ) cabal: Error: some packages failed to install: happstack-util-0.5.0 failed during the building phase. The exception was: ExitFailure (-1073741819) A Windows prompt shows problems (Application not properly initialized) with a "perl.exe" program. Does cabal use perl (that's completely unexpected for me)? I did install perl and put it on the path and that didn't solve it (maybe there is a configuration parameter to point to perl's executable?).... Thanks, Flavio

4 4

DTP10 Call for Participation
by Conor McBride 06 May '10

06 May '10

Remember, Haskell is the world's most popular dependently typed functional programming language... (s:S)*(p:P s)->(s:S)*(p:P s)->(s:S)*(p:P s)->(s:S)*(p:P s)->(s:S)*(p:P s)-> DTP 2010 --- Call for Participation EARLY REGISTRATION ENDS 17 MAY 2010 Workshop on DEPENDENTLY TYPED PROGRAMMING Edinburgh, Scotland, 9&10 July 2010 (a FLoC workshop, affiliated with LICS) http://sneezy.cs.nott.ac.uk/darcs/dtp10/ (s:S)*(p:P s)->(s:S)*(p:P s)->(s:S)*(p:P s)->(s:S)*(p:P s)->(s:S)*(p:P s)-> Roll up! Roll up! Register early, register often! http://www.floc-conference.org/registration.html Attendance at DTP10 can be yours at a BARGAIN price if you register BEFORE 17 MAY 2010. The preliminary programme for DTP10 is here: http://sneezy.cs.nott.ac.uk/darcs/dtp10/programme.html Invited Talks: Ana Bove, Chalmers University, "10 Years of Partiality and General Recursion" Matthieu Sozeau, Harvard University, "Elaborations in Type Theory" Contributed Talks: Edwin Brady, "Practical, efficient programming with dependent types" James Caldwell, "Extracting Monadic Programs form Proofs", (joint work with Josef Pohl) Adam Chlipala, "Generating Pieces of Web Applications with Type-Level Programming" Nils Anders Danielsson, TBA Larry Diehl, "Unit & integration test composition via lemmas" Makoto Hamana, "Another Initial Algebra Semantics of Inductive Families for Programming" Hugo Herbelin, "A sequent calculus presentation of the Calculus of Inductive Constructions" (joint work with Jeffrey Sarnat, Vincent Siles) Karim Kariso, "Integrating Agda and Automated Theorem Proving Techniques" (joint work with Anton Setzer) Dan Licata, "Security-Typed Programming within Dependently Typed Programming" (joint work with Jamie Morgenstern) Ulf Norell, TBA Carsten Schuermann, "The HOL-Nuprl connection in Delphin", (joint work with Adam Poswolsky) Anton Setzer, "Coalgebras in dependent type theory" Antonis Stampoulis, "VeriML: Type-safe computation of logical terms inside a language with effects" Tarmo Uustalu, TBA Sean Wilson, "Supporting Dependently Typed Functional Programming with Proof Automation and Testing" If, by some chance, you are interested in talking at DTP10, please do get in touch. Space in the programme is now very tight, but we remain open to proposals. See you in Edinburgh in July! Thorsten and Conor _______________________________________________ Agda mailing list Agda(a)lists.chalmers.se https://lists.chalmers.se/mailman/listinfo/agda

2 2

Interest in a Mathematics & AI strike force ?
by Alp Mestanogullari 06 May '10

06 May '10

Hello -cafe, When I started learning Haskell, I saw the AI page [1] which aimed at creating a sound, uniform and handy framework for AI programming in Haskell. I added my name on it and thought a bit about it. I even wrote a first version of HNN [2], a neural network library, quite early in my Haskell days. I found that idea to be great but did not see any actual effort around this. So, I'm now thinking again about that and even enlarging it to mathematics & AI. Thus, I would like to have an idea of the number of people interested in being involved in such an effort. There are several tools out there on hackage but they aren't that much uniform and neither play nicely together. I'm pretty convinced this could be improved and as a Mathematics student I'm highly interested in that. If enough people are interested, we could for example set up a mailing list and a trac to organize the effort and then people could just discuss and write Haskell modules when time permits. Any comment, idea, reaction, interest ? [1] http://www.haskell.org/haskellwiki/AI [2] http://www.haskell.org/haskellwiki/HNN -- Alp Mestanogullari http://alpmestan.wordpress.com/ http://alp.developpez.com/

6 16

Mystery Parse Error in LHS file
by Iæfai 06 May '10

06 May '10

I have an lhs file, 'starsystem.lhs' that is not compiling because of a parse error. Specifically: $ cabal build Preprocessing executables for starsystem-0.1... Building starsystem-0.1... [1 of 4] Compiling Render ( Render.lhs, dist/build/ starsystem/starsystem-tmp/Render.o ) [2 of 4] Compiling Passive ( Passive.lhs, dist/build/ starsystem/starsystem-tmp/Passive.o ) [3 of 4] Compiling Active ( Active.lhs, dist/build/ starsystem/starsystem-tmp/Active.o ) [4 of 4] Compiling Main ( starsystem.lhs, dist/build/ starsystem/starsystem-tmp/Main.o ) starsystem.lhs:29:0: parse error on input `main' The location is at the first instance of main: > data Action = Action (IO Action) The main program is mostly book-keeping such as initializing OpenGL and GLFW, creating window, setting up viewport, etc. \begin{code} main :: IO () -- ** line 29 main = do c_unbundled GLFW.initialize -- open window I cannot figure out what the problem here is. I have tried it in a regular hs file and it works. I have posted full source here: http://hpaste.org/fastcgi/hpaste.fcgi/view?id=25352#a25352 Any thoughts would be welcome.

3 5

ANNOUNCE: wai-handler-fastcgi 0.0.0
by Michael Snoyman 06 May '10

06 May '10

Hi all, I finally scratched an itch I've had for a while, and put together wai-handler-fastcgi. It is built on top of Dan Knapp's wonderful direct-fastcgi[1] package, so it is free from dependencies on C libraries. This package allows WAI[2] applications to be run on a FastCGI-supporting server. The WAI, for those not familiar, is a standard interface between web servers and web applications, allowing you to write your web applications once and deploy in multiple manners. Michael [1] http://hackage.haskell.org/package/direct-fastcgi [2] http://hackage.haskell.org/package/wai

1 0