- Haskell - Haskell.org

Or-Patterns
by Lajos Nagy 26 Oct '06

26 Oct '06

Whatever happened to the idea of Or-Patterns? http://www.haskell.org/pipermail/glasgow-haskell-users/2000-December/001535… I find them more than occasionally useful :) Is there some theoretical difficulty or is it just that nobody had the time/will to implement it? Regards, Lajos Nagy

1 0

Monad transformer question
by Cyril Schmidt 26 Oct '06

26 Oct '06

Working on a Monte-Carlo simulation where I have to calculate the values of a certain function on the given set of inputs, I noticed that some of the input variables change for every iteration, while others do not. To give a simple example, let's suppose I have a function f a1 a2 p = a1*a2 + p and I have to get its values for [ (a1,a2,p) | a1 <- [0.1,0.2], a2 <- [0.1,0.2], p <- [0..9] ] For efficiency, I want to pre-calculate (a1*a2) for each pair of a1 and a2, and then calculate f for each p. (The real function is far more complicated, but the idea is the same: pre-calculate all that depends on seldom-changing variables, and then run the rest of the iterations using the pre-calculated value). I built my function f as a Reader monad because my real f has many input parameters, so it is handy to have them packed in the Reader's environment. There are two Readers, in fact: the first environment contains the seldom- -changing variables, the second contains the variable that changes often. My code looked as follows. >{-# OPTIONS -fno-monomorphism-restriction #-} >module Main where >import Control.Monad.Reader The variable that changes most often: >data Inner = Inner { p1 :: Double } The variables that change seldom >data Outer = Outer { a1 :: Double, > a2 :: Double } Function precalc pre-calculates (a1*a2) >precalc = do > a1 <- asks a1 > a2 <- asks a2 > let r = {-# SCC "r" #-} a1*a2 > return r Function f in monadic form: >f :: Reader Outer (Reader Inner Double) >f = do > r <- precalc > return $ do { p1 <- asks p1 > ; let s = {-# SCC "s" #-} r+p1 > ; return s } Function runf runs f over all values of p: >runf (a1,a2) = do > let reader = runReader f $ Outer a1 a2 > results = map (runReader reader) [ Inner { p1 = x } | x <- [0..9] ] > putStrLn $ "a1 = "++show a1++", a2= "++show a2++", results = " > ++show results The main function >main = mapM_ runf [(a1,a2) | a1 <- [0.1,0.2], a2 <- [0.1,0.2]] This all works fine; the profiler shows that the (a1*a2) calculation is performed exactly 4 times, while addition, just as expected, 40 times. I noticed that f f :: Reader Outer (Reader Inner Double) can be implemented using monad transformer: f' :: ReaderT Outer (Reader Inner) Double The only difference in the implementation is that f' uses lift instead of return: >f' :: ReaderT Outer (Reader Inner) Double >f' = do > r <- precalc > lift $ do { p1 <- asks p1 > ; let s = {-# SCC "s" #-} r+p1 > ; return s } > >runf' (a1,a2) = do > let reader = runReaderT f' $ Outer a1 a2 > results = map (runReader reader) [ Inner { p1 = x } | x <- [0..9] ] > putStrLn $ "a1 = "++show a1++", a2= "++show a2++", results = " > ++show results However similar they look, f and f' have very different behaviour (their results are the same, of course). When I use runf' instead of runf, the profiler shows that precalc is invoked 40 times, which means that all the benefits of pre-calculating (a1*a2) are gone. (In the real application, I pre-calculate a much more complicated and expensive expression, that's why it matters). I am curious why this happens. As far as I can see, the lift function of ReaderT is the same as return of Reader, and the >>= in Reader and ReaderT are pretty similar to each other, so why is the behaviour different? This is a question of a purely theoretical significance for me; it does not hinder my work in any way. Still, I would greatly appreciate any ideas. By the way, I am using GHC 6.4.2 on Windows. Kind regards, Cyril

3 2

package mounting
by Frederik Eaton 25 Oct '06

25 Oct '06

Hi lists, I recently read Simon Peyton Jones' proposal: http://hackage.haskell.org/trac/ghc/wiki/GhcPackages and disagreed with some of the design decisions. (To be fair, the aspects I disagree with are shared with most or all of the other proposals) So I've put an alternative proposal here: http://hackage.haskell.org/trac/ghc/wiki/PackageMounting Discussion is welcome (but only on the libraries@ list, please). Hopefully we can weigh the benefits of each proposal, and reach an agreement and write up a more formal specification within the next few months or so provided that people decide that this is an extension we want to implement. Cheers, Frederik P.S. (I'm sorry that I missed Sven's earlier package mounting thread in July. But I hope the above article covers what my response would have been) -- http://ofb.net/~frederik/

1 0

Re: [Haskell] Monad transformer question
by Cyril Schmidt 25 Oct '06

25 Oct '06

Lajos, Point taken, thanks. My question, though, is not as much about how I speed up my program, as it is about the behaviour of the ReaderT monad transformer. There is something I do not understand about it, namely, why does such a small change in the code change the behaviour so much. Regards, Cyril > Based on what you wrote, I'm not sure that monads are the right answer > in this situation. It seems like to me that simple memoization of > function > values would yield a purely functional but equally efficient solution. > > A typical way to memoize the fibonacci function: > > fib = ((map fib' [0 ..]) !!) > where > fib' 0 = 0 > fib' 1 = 1 > fib' n = fib (n - 1) + fib (n - 2) > > (It only gives you a speed up if you compile with -O2 though.) > > Regards, > > Lajos Nagy

1 0

Compilation of big, computed tables
by Stefan Karrmann 24 Oct '06

24 Oct '06

Dear all, can ghc compile huge tables into efficient code if they are constant at compile time? Two examples may clearify the question: big1 :: UArray Int Char big1 = array (0,1000) $! map (\i -> (i,toEnum i)) [0..1000] big2 = sum [0..10000]::Int -- == 50005000 == n*(n+1)/2 where n = 10000 Both values are constant at compile time. As they are given by pure functions, the compiler could evaluate them and write the *result* into the object file 'foo.o'. This would save code size and run time. I peeked into 'foo.hc' but I didn't found 0x2fb0408 nor an array {0, 1, 2, 3, ..., 1000} or similar things. The function in big2 should show that the computation of the value could be very time consuming. If the compiler does not compute it, the source file could be generated by a helper program (or template Haskell?). The function in big1 should show that the conversion of the data into an array could be run-time, code-size and heap-size consuming. (You need the list (maybe explicitly - think of 1000 fixed pseudo-random numbers) and convert it into an array.) If the compiler generates the unboxed array directly it could be rather efficient. PS: I compiled with: ghc6 -c -O2 -keep-tmp-files -keep-hc-files foo.hs Regards, -- Stefan Karrmann

9 12

Haskell Weekly News: October 24, 2006
by dons＠cse.unsw.edu.au 23 Oct '06

23 Oct '06

--------------------------------------------------------------------------- Haskell Weekly News http://haskell.org/haskellwiki/HWN Issue 46 - October 24, 2006 --------------------------------------------------------------------------- Welcome to issue 46 of HWN, a weekly newsletter covering developments in the Haskell community. Announcements * MissingH 0.16.0. John Goerzen [1]announced that the latest version of MissingH is now available. MissingH is a suite of 'missing' library functions. New features include: render numbers as binary units, a progress tracker, turn QuickCheck tests into HUnit tests, and GHC 6.6 support. 1. http://article.gmane.org/gmane.comp.lang.haskell.general/14400 * SMP parallel Pugs on GHC. Audrey Tang [2]announced that parallel support, on top of GHC's new SMP runtime system, has been added to Pugs, the standard bearer [3]Perl6 implementation. 2. http://thread.gmane.org/gmane.comp.lang.haskell.general/14402/focus=14402 3. http://pugs.blogs.com/pugs/2006/10/smp_paralleliza.html * YAHT is now a part of the wikibook. Eric Kow [4]announced that the famous 'Yet Another Haskell Tutorial' has been imported into [5]the Haskell wikibook. Let the great Haskell Remix begin! 4. http://koweycode.blogspot.com/2006/10/yaht-badly-imported.html 5. http://en.wikibooks.org/wiki/Haskell Haskell' This section covers the [6]Haskell' standardisation process. * [7]Pattern guards * [8]Standard syntax for preconditions, postconditions, and invariants * [9]Indentation of If-Then-Else * [10]Module imports anywhere 6. http://hackage.haskell.org/trac/haskell-prime 7. http://thread.gmane.org/gmane.comp.lang.haskell.prime/1750/focus=1750 8. http://thread.gmane.org/gmane.comp.lang.haskell.prime/1773/focus=1773 9. http://thread.gmane.org/gmane.comp.lang.haskell.prime/1763/focus=1763 10. http://thread.gmane.org/gmane.comp.lang.haskell.prime/1764/focus=1764 Discussion * Ruby puzzles. Jim Burton [11]mentioned that he was working on the [12]Ruby quiz puzzles in Haskell -- an interesting exercise. 11. http://thread.gmane.org/gmane.comp.lang.haskell.cafe/16056/focus=16056 12. http://www.rubyquiz.com/ * DeepSeq and parallel strategies. Chad Scherrer [13]started a bit of a discussion about the connection between the deepSeq function, and Control.Parallel.Strategies.rnf 13. http://thread.gmane.org/gmane.comp.lang.haskell.cafe/16047/focus=16047 * Type level functions. Oleg Kiselyov [14]described how to use type level programming to create a type of constrained lists. 14. http://article.gmane.org/gmane.comp.lang.haskell.cafe/15978 * Extending the core libraries. Josef Svenningsson [15]started a largish thread on adding a concat.intersperse function to the base library, leading to an interesting discussion on whether the time has come for an formal process for extending core libraries. 15. http://thread.gmane.org/gmane.comp.lang.haskell.libraries/5280/focus=5280 * data Void. Conor McBride [16]proposed adding the type with no inhabitants other than _|_ to the core libraries. 16. http://thread.gmane.org/gmane.comp.lang.haskell.libraries/5321/focus=5321 Blog noise [17]Haskell news from the blogosphere. * [18]Optimus prime is actually dead, or, monadic transformers * [19]Haskell, PDF and penrose tilings * [20]Not everyone can dine at the Beverly Hills Social Club of Programming Languages * [21]Would you like a side of referential transparency with your order.. * [22]John Baez on lambda calculus and games * [23]Monads: a field guide * [24]Catch+YHC beats GHC 17. http://planet.haskell.org/ 18. http://brokenhut.livejournal.com/172054.html 19. http://www.alpheccar.org/en/posts/show/57 20. http://sequence.complete.org/node/215 21. http://weblog.raganwald.com/2006/10/would-you-like-side-of-referential.html 22. http://wadler.blogspot.com/2006/10/john-baez-on-lambda-calculus-and-games.h… 23. http://sigfpe.blogspot.com/2006/10/monads-field-guide.html 24. http://neilmitchell.blogspot.com/2006/10/30-faster-than-ghc.html About the Haskell Weekly News Each week, new editions are posted to [25]the Haskell mailing list as well as to [26]the Haskell Sequence and [27]Planet Haskell. [28]RSS is also available, and headlines appear on [29]haskell.org. The Haskell Weekly News is also [30]available in Spanish translation. To help create new editions of this newsletter, please see the [31]contributing information. Send stories to dons at cse.unsw.edu.au. The darcs repository is available at darcs get http://www.cse.unsw.edu.au/~dons/code/hwn 25. http://www.haskell.org/mailman/listinfo/haskell 26. http://sequence.complete.org/ 27. http://planet.haskell.org/ 28. http://sequence.complete.org/node/feed 29. http://haskell.org/ 30. http://haskell.org/haskellwiki/HWN/es 31. http://haskell.org/haskellwiki/HWN

1 0

Re: Haskell Weekly News: September 27, 2006
by Cyril Schmidt 23 Oct '06

23 Oct '06

Donald Bruce Stewart wrote: >Conference roundup > The [41]Commercial Users of Functional Programming workshop (CUFP), > held in association with [42]ICFP, attracted a record turn-out this > year. [...] > * Haskell programming at Credit Suisse. Howard Mansell talked about > Haskell programming at [44]Credit Suisse in New York. Sadly, I was unable to attend. Is the presentation, by any chance, available online? Pointers to other related info will be much appreciated. (Yes, I know that a report of the conference will be published, but that won't happen before December :( Regards, Cyril

4 5

Pugs gains SMP parallelism support.
by Audrey Tang 23 Oct '06

23 Oct '06

I hacked +RTS -N support into Pugs today; here's a short writeup: http://pugs.blogs.com/pugs/2006/10/smp_paralleliza.html Pugs's current implementation for concurrent operations on lists is very naive: chan <- newChan forM ([0..] `zip` xs) $ \(n, x) -> forkIO $ do rv <- runEvalIO env (reduce x) writeChan chan (n, rv) fmap (map snd . sort) (replicateM (length xs) (readChan chan)) While the initial result on Linux 2.6 is encouraging, on OSX/Intel with two CPUs it actually slightly slows down the program when running on -N2 or above. I wonder if there is a more efficient way doing this... Thanks, Audrey

5 12

ANN: MissingH 0.16.0
by John Goerzen 19 Oct '06

19 Oct '06

Hello, The latest version of MissingH is now up at http://quux.org/devel/missingh The changes for this version are: * NEW: MissingH.Quantity that is used to render numbers as binary units (MB, KB, etc) or SI units, or other similar units * NEW: MissingH.ProgressTracker that is used to track the progress of long-running tasks, measure their speed, and give estimated completion times * Update to support GHC 6.6 + Remove FiniteMap support everywhere (no longer supported in GHC) + Remove MissingH.Printf (has been deprecated since 0.13.0 due to Text.Printf) + Update README + Fix MissingH.FTP.Server for GHC 6.6.0 compatibility + Update cabal file for GHC 6.6 compatibility * Patch from Jeremy Shaw to add to MissingH.HUnit the ability to turn QuickCheck tests into HUnit tests * Patch from Joe Edmonds to MissingH.Wash.Mail.MailParser to accept a slight violation of RFC2822 in which date headers may have a single-digit date * Fix a logic error in MemoryBuffer's vPutStr that could corrupt data when a write occurs after a seek, and wrote a test to detect this condition. Thanks to Bulat Ziganshin for finding this. * New function in MissingH.List to merge two sorted lists into a single sorted whole. Patch from Clifford Beshers. * Work around lack of signals on Windows in MissingH.Cmd. * Make sure Data.Bits is imported in MissingH.IO.StatCompat. -- John Goerzen Author, Foundations of Python Network Programming http://www.amazon.com/exec/obidos/tg/detail/-/1590593715

1 0

cabal/haddock
by Johannes Waldmann 19 Oct '06

19 Oct '06

Hello. What's the current state of Haddock support in Cabal? I checked that "runhaskell Setup.hs haddock" works fine, but how can I reference the documentation of (cabalized) package Y from the docs of package X? - Last I asked this, the answer was "write your hooks in Setup.hs" but I was hoping that meanwhile I could just put the info in the .cabal file. -- -- Johannes Waldmann -- Tel/Fax (0341) 3076 6479/80 -- ---- http://www.imn.htwk-leipzig.de/~waldmann/ -------

2 1