March 2012 - Haskell-Cafe

ANNOUNCE: haskell-src-exts-1.12.0
by Niklas Broberg 27 Mar '12

27 Mar '12

Fellow Haskelleers, I'm pleased to announce the release of haskell-src-exts-1.12.0! * On hackage: http://hackage.haskell.org/package/haskell-src-exts * Via cabal: cabal install haskell-src-exts * Darcs repo: http://code.haskell.org/haskell-src-exts It's been a long while since the last update before this, but it is my definite hope that I will once again be able to update regularly. This major release primarily enables the (relatively) new quasi-quoter syntax. The "major" version bump is questionable, but better safe than sorry. The only backwards incompatible change is that pretty- and exact-printing now use the new quasi-quoter syntax, which makes the round-tripping 'exactPrint . parse' cycle fail for code using the old syntax. There are more changes in the pipeline, some larger and some smaller. Stay tuned! Changelog: 1.11.1 --> 1.12.0 =============== * Move from old [$...| quasi-quote syntax to the new [...| one. The old syntax is still recognized while parsing. * Allow symbols as variables when TypeOperators is enabled. * Rename ExplicitForall in ExplicitForAll, to be consistent with GHC and the Haskell' process. Cheers, /Niklas

2 1

Reminder Munich Haskell Meeting
by Heinrich Hördegen 27 Mar '12

27 Mar '12

Dear all, just to remind you: Tomorrow, 28th of March, our Haskell get-together will take place in Munich. If you plan to join, go to http://www.haskell-munich.de/dates and click the button. Have a nice day, Heinrich -- -- hoerdegen(a)funktional.info www.funktional.info --

1 0

Regarding Haskell FFI for C/C++ (Passing of String)
by rajendra prasad 27 Mar '12

27 Mar '12

Hi, I have just started learning Haskell FFI. I am trying to send a string from hastell to a C function. For this, I am required to convert the haskell string to byte string. I have two methods to achieve this task. Both are listed below: 1) import Foreign.C.String let arg1 = map castCharToCChar "Hello" :: [CChar] 2) import qualified Data.ByteString.Char8 as B f = B.pack "Hello" I just wanted to know the optimal way to achieve this task. Please suggest the optimal way of doing this. If there is any other way, please share it. Also, please suggest me any good tutorial to start with Haskell FFI for C/C++. Thank you very much. Regards, Rajendra

2 3

good lightweight web-framework like sinatra?
by serialhex 27 Mar '12

27 Mar '12

i'm looking for something lightweight, that dosnt need it's own server, can easily run on cgi on an apache with minimal work, and dosn't have many dependancies. i was looking at yesod, but it is bigger than i need for my site (at this point) and would take too much work to get running on my webhost. though i am looking forward to learning it and using it in the future, i just need something that will play nicely with apache & cgi... justin p.s. if anyone is interested to know i'm using nearlyfreespeach.net as my host... haskell is one of the many languages they support via cgi, but, at the moment, it is kind of difficult to get yesod or rails or the like to work on it... :-/ -- * The wise man said: "Never argue with an idiot. They bring you down to their level and beat you with experience." * As a programmer, it is your job to put yourself out of business. What you do today can be automated tomorrow. ~Doug McIlroy No snowflake in an avalanche ever feels responsible. --- CFO: “What happens if we train people and they leave?” CTO: “What if we don’t and they stay?”

6 6

Is there a generic way to detect "mzero"?
by Ting Lei 27 Mar '12

27 Mar '12

Hi, I was writing a code trying to use MonadPlus to detect some error cases (representing missing values etc. in pure code). With the Maybe monad, I can do this: can0 :: (a -> Maybe b) -> a -> Bool can0 f x = case f x of Nothing -> False Just x -> True And I got the expected result: *Main> can0 (\x -> Just x) 1 True But, when I try to generalize this using MonadPlus, as follows: can :: (MonadPlus m) => (a -> m b) -> a -> Bool can f x = case f x of mzero -> False _ -> True I got a warning: __testError.hs:31:11: Warning: Pattern match(es) are overlapped In a case alternative: _ -> ... Ok, modules loaded: Main. And the result is also not as intended (see also can0): *Main> can (\x -> Just x) 1 False Can anyone help to explain why this wouldn't work or if there is a workaround to use Monadplus and mzero (or Monad and fail) to achieve this? Thanks in advance for your help Ting

9 10

Need a backup GSOC to apply for? Last minute new proposal -- memory reuse analysis for GHC
by Ryan Newton 26 Mar '12

26 Mar '12

Hi potential GSOC'ers, If there are multiple students interested in the project you're applying to it's a good idea to put in more than one application. This is a project proposal that would focus on performance analysis -- in particular in reusing some of the tools for analyzing memory access patterns in the context of Haskell: http://hackage.haskell.org/trac/summer-of-code/ticket/1619 I think this project would yield pretty good bang for the buck because almost all the pieces are already there. They just need to be combined, polished, documented, and evangelized. Cheers, -Ryan

1 0

ANN: Latest package versions on Hackage as JSON/JSONP
by Simon Hengel 26 Mar '12

26 Mar '12

Hi, I provide a relation between /package name/ and /latest version/ on Hackage as both JSON and JSONP. http://www.typeful.net/~tbot/hackage/ This are static files, and they are regenerated whenever new packages are uploaded (with a delay of about one minute). Two simple usage examples are given in the README[1]. Feel free to access this information for whatever purpose! Cheers, Simon [1] https://github.com/sol/hackage-jsonp#readme

2 1

difficulty building (some) TH packages with ghc-7.4.1
by John Lato 26 Mar '12

26 Mar '12

Hello, I've run into an odd problem when building certain packages that use Template Haskell with GHC-7.4.1. For example, RepLib: $ cabal --version cabal-install version 0.13.3 using version 1.14.0 of the Cabal library $ cabal install -O2 -w ~/.ghc-7.4.1/bin/ghc RepLib --reinstall Resolving dependencies... Configuring RepLib-0.5.2... Building RepLib-0.5.2... Preprocessing library RepLib-0.5.2... [ 1 of 12] Compiling Generics.RepLib.R ( Generics/RepLib/R.hs, dist/build/Generics/RepLib/R.o ) [ 2 of 12] Compiling Generics.RepLib.R1 ( Generics/RepLib/R1.hs, dist/build/Generics/RepLib/R1.o ) [ 3 of 12] Compiling Generics.RepLib.Derive ( Generics/RepLib/Derive.hs, dist/build/Generics/RepLib/Derive.o ) [ 4 of 12] Compiling Generics.RepLib.PreludeReps ( Generics/RepLib/PreludeReps.hs, dist/build/Generics/RepLib/PreludeReps.o ) Loading package ghc-prim ... linking ... done. Loading package integer-gmp ... linking ... done. Loading package base ... linking ... done. Loading package type-equality-0.1.0.2 ... linking ... done. Loading package array-0.4.0.0 ... linking ... done. Loading package deepseq-1.3.0.0 ... linking ... done. Loading package containers-0.4.2.1 ... linking ... done. Loading package pretty-1.1.1.0 ... linking ... done. Loading package template-haskell ... linking ... done. Loading package transformers-0.2.2.0 ... linking ... done. Loading package mtl-2.0.1.0 ... linking ... done. cabal: Error: some packages failed to install: RepLib-0.5.2 failed during the building phase. The exception was: ExitFailure 11 This is without library profiling enabled. If I enable library profiling, the unprofiled build is fine but the profiled build exits with the same error, at the same location. If I pass the "-v" argument to cabal-install, the package builds fine with profiling disabled. The "-v" flag makes no difference with profiling enabled, although some packages (but not RepLib) work with "-v2" I haven't seen this with ghc-7.2.*, but I'm guessing this is some sort of cabal-install bug. Has anyone else witnessed anything like this? Thanks, John L.

1 0

Using HaXml
by Yves Parès 26 Mar '12

26 Mar '12

Hello café, Provided what I read, HaXml seems to be the recommended library for parsing XML files (I'm trying to browse and alter spreadsheets (ODS) and possibly release a package when I'm done), is there somewhere tutorials on how to use it? I used 'xml' package by the past, but HaXml is more substantial. Thanks.

2 2

Execute only one step in Hood-Melville Real time queue.
by Xinyu LIU 26 Mar '12

26 Mar '12

Hi, I read the Hood-Melville real time queue realization in [1]. There are 2 lists maintained in queue, front and rear. When the queue gets unbalanced due to push/pop, it amortized the f ++ reverse r incrementally to build the new front list based on the below mechanism reverse r = reverse' r [] where reverse' [] acc = acc reverse' (x:xs) = reverse' xs (x:acc) and f ++ reverse r == reverse (reverse f) ++ reverse r == reverse' (reverse f) [] ++ reverse r == reverse' (reverse' f []) (reverse' r []) And the re-balance happens once | f | + 1 = | r |. Let's denote m = |f|. incremental f ++ reverse r takes total 2*m + 2 steps. In the the realization mentioned in [1], it execute 2 steps every push and pop to make sure the incremental computation finish before next queue re-balancing. However, I found it's possible to execute only one step per push/pop. because: 1. Next re-balance happens at earliest | f' | + 1 = |r|+|f|+1+1 = 2*m + 2 times by continuously push operation; 2. If we keep a copy of f, and a counter of how many elements left in f, which need to be incrementally 'appended', by continuously m times popping operation. we can finish the f ++ reverse r (actually, only reverse r is needed, as all elements in f are popped) Based on this fact, I rewrite the program as the following: data State a = Empty | Reverse Int [a] [a] [a] [a] -- n, f', acc_f' r, acc_r | Append Int [a] [a] -- n, rev_f', acc | Done [a] -- result: f ++ reverse r deriving (Show, Eq) -- front, length of front, on-goint reverse state, rear, length of reverse data RealtimeQueue a = RTQ [a] Int (State a) [a] Int deriving (Show, Eq) -- we skip the empty error for pop and front empty = RTQ [] 0 Empty [] 0 isEmpty (RTQ _ lenf _ _ _) = lenf == 0 -- O(1) time push push (RTQ f lenf s r lenr) x = balance f lenf s (x:r) (lenr + 1) -- O(1) time pop pop (RTQ (_:f) lenf s r lenr) = balance f (lenf - 1) (abort s) r lenr front (RTQ (x:_) _ _ _ _) = x balance f lenf s r lenr | lenr <= lenf = step f lenf s r lenr | otherwise = step f (lenf + lenr) (Reverse 0 f [] r []) [] 0 -- execute f ++ reverse r step by step step f lenf s r lenr = case s' of Done f' -> RTQ f' lenf Empty r lenr s' -> RTQ f lenf s' r lenr where s' = if null f then next $ next s else next s next (Reverse n (x:f) f' (y:r) r') = Reverse (n+1) f (x:f') r (y:r') next (Reverse n [] f' [y] r') = Append n f' (y:r') next (Append 0 _ acc) = Done acc next (Append n (x:f') acc) = Append (n-1) f' (x:acc) next s = s -- Abort unnecessary appending as the element is popped abort (Append 0 _ (_:acc)) = Done acc -- Note! we rollback 1 elem abort (Append n f' acc) = Append (n-1) f' acc abort (Reverse n f f' r r') = Reverse (n-1) f f' r r' abort s = s Note the 'where' clause in step function. This is because we need an extra step to change the state from (Append 0 _ xs) to (Done xs). I tested this program with invariant testing with QuickCheck. The behavior is correct as other queue implementation. The program can be found here: https://github.com/liuxinyu95/AlgoXY/blob/algoxy/datastruct/elementary/queu… Reference: [1]. Chris Okasaki. ``Purely functional data structures''. P102. Section 8.2.1. Cambridge University Press. ISBN 0521663504 -- Larry, LIU Xinyu https://sites.google.com/site/algoxy/ https://github.com/liuxinyu95/AlgoXY

1 0