- Haskell-Cafe - Haskell.org

Typeable and fancy types
by Ron Alford 11 Jul '08

11 Jul '08

Hey all, I've reduced my previous problem to a small example. Anyone know why typeOf (...) would work, but typeOf [...] would not? Is the derivation for lists funky? data Expr f = In (f (Expr f)) instance Typeable1 f => Typeable (Expr f) where typeOf (In x) = mkTyConApp (mkTyCon "TypeTest.Expr") [typeOf1 x] data Foo e = Foo deriving instance Typeable1 Foo *TypeTest> typeOf (In Foo) TypeTest.Expr TypeTest.Foo *TypeTest> typeOf [In Foo] *** Exception: Prelude.undefined

3 2

Newbie: Appending arrays?
by Dmitri O.Kondratiev 11 Jul '08

11 Jul '08

What is the best way to extend array? I would use a list instead of array as it is easy to append, but need to have random access to its elements later. So in fact I need to start with an integer array of size 1. Next I may need to add new elements to this array or modify values of the existing ones. Function: array :: (Ix a) => (a,a) -> [(a,b)] -> Array a b allows construct an array of a fixed size. How to add more elements to the array later? Thanks! -- Dmitri O. Kondratiev dokondr(a)gmail.com http://www.geocities.com/dkondr

7 11

Inductive graphs memory usage
by Andre Nathan 11 Jul '08

11 Jul '08

Hello I'm trying to create a directed graph using the Data.Graph.Inductive. The graph is a random graph using the G(n, p) model, that is, each of the n nodes is linked to every other node with probability p. I'm seeing a large increase of memory usage when n grows (this is using p = 0.1): n = 1000 -> 96MB n = 2000 -> 283MB n = 3000 -> 760MB So, I'm probably doing something very stupid :) The code is below. Is there anything I could do to optimize memory usage here? module Main where import Control.Monad import Data.Graph.Inductive import System.Random createEdges :: Int -> Double -> IO [LEdge Int] createEdges n prob = foldM create [] [1..n] where create es i = foldM (flip $ link i) es [i, i-1 .. 1] link i j es | i == j = return es -- no self-loops | otherwise = do es' <- maybeCreateEdge i j prob es es'' <- maybeCreateEdge j i prob es' return es'' maybeCreateEdge :: Node -> Node -> Double -> [LEdge Int] -> IO [LEdge Int] maybeCreateEdge i j prob es = do r <- randomDouble return $ if r < prob then (i, j, 0):es else es randomDouble :: IO Double randomDouble = getStdRandom $ random main :: IO () main = do let (n, p) = (3000, 0.1) es <- createEdges n p let g = mkGraph [(i, 0) | i <- [1..n]] es :: Gr Int Int return () Thanks, Andre

4 7

ANN: Takusen 0.8.3
by Alistair Bayley 11 Jul '08

11 Jul '08

Changes since 0.8.1 (I put 0.8.2 up on hackage with an error in Setup.hs, so it's been skipped): - ODBC support: datetime marshalling is improved. For bind parameters this uses the timestamp struct for most back-ends, but String for MS SQL Server because populating the timestamp struct always failed. - more Cabal improvements: now uses configurations, so the Setup.hs script should be both simpler and more robust. Requires Cabal >= 1.4. Oracle backend on Linux should build nicely. - bug fix for a resource leak if an exception was thrown when initiating a query (the Statement handle was not closed). - some basic result-set validation against the iteratee: if you try to fetch a column that is not in the result-set, an error is thrown (rather than garbage returned). The release bundle: http://hackage.haskell.org/packages/archive/Takusen/0.8.3/Takusen-0.8.3.tar… The latest code: darcs get http://darcs.haskell.org/takusen Docs: http://darcs.haskell.org/takusen/doc/html/index.html A comprehensive description of API usage can be found in the documentation for module Database.Enumerator (look for the Usage section): http://darcs.haskell.org/takusen/doc/html/Database-Enumerator.html Future plans: - Output bind-parameters and multiple-result sets for ODBC - FreeTDS backend (Sybase and MS Sql Server) - support for Blobs and Clobs For those of you unfamiliar with Takusen, here is our HCAR blurb: Takusen is a DBMS access library. Like HSQL and HDBC, we support arbitrary SQL statements (currently strings, extensible to anything that can be converted to a string). Takusen's 'unique-selling-point' is safety and efficiency. We statically ensure all acquired database resources - such as cursors, connection and statement handles - are released, exactly once, at predictable times. Takusen can avoid loading the whole result set in memory, and so can handle queries returning millions of rows in constant space. Takusen also supports automatic marshalling and unmarshalling of results and query parameters. These benefits come from the design of query result processing around a left-fold enumerator. Currently we fully support ODBC, Oracle, Sqlite, and PostgreSQL.

2 1

Combining Wouter's expressions with extensible records
by Ron Alford 11 Jul '08

11 Jul '08

Well, my extension of Wouter's datatypes proved to be unweildy.... So, I'm trying to use http://fmapfixreturn.wordpress.com/2008/05/03/simple-extensible-records-now… for extensible records. I ran across my first problem rather quickly! data Expr f = In (f (Expr f)) Ok, but to make it part of a record, it needs to implement Data: data Expr f = In (f (Expr f)) deriving Data but this gives No instances for (Data (f (Expr f)), Typeable (Expr f)) arising from the 'deriving' clause of a data type declaration at Planning/Wouter.hs:77:0-42 Any hints? Thanks, -Ron

2 12

More idiomatic use of strictness
by Grzegorz Chrupala 11 Jul '08

11 Jul '08

Hi all, Is there a less ugly way of avoiding laziness in the code pasted below then the use of seq in the last line? The program is supposed to split a large input file into chunks and check in how many of those chunks each of a list of words appear, as well as the total number of chunks. Without the seq it consumes huge amounts of memory. Thanks! Grzegorz {-# LANGUAGE BangPatterns, PatternGuards #-} import Data.List (foldl') split delim s | [] <- rest = [token] | otherwise = token : split delim (tail rest) where (token,rest) = span (/=delim) s main = do putStrLn =<< fmap (show . stats ["the","a","and"] . split "<DOC>" . words) getContents stats ws docs = foldl' f ((map (const 0) ws),0) docs where f (dfs,n) d = let dfs' = zipWith (\w df -> (df + fromEnum (w `elem` d))) ws dfs in sum dfs' `seq` (dfs',n+1) -- View this message in context: http://www.nabble.com/More-idiomatic-use-of-strictness-tp18379800p18379800.… Sent from the Haskell - Haskell-Cafe mailing list archive at Nabble.com.

3 3

FPGA / Lava and haskell
by Satnam Singh 11 Jul '08

11 Jul '08

Hello Marc. I've been distracted with the Kiwi project (parallel C# programs to hardware) and right now a Bluespec project so I've not made any progress on a merge of the Lava I produced for designing circuits (with layout) for Xilinx's FPGAs and the Chalmers version (which I recommend for situations where you don't care about circuit layout or when you don't want to use FPGA architecture specific features). The version of Lava on my website at http://raintown.org/lava/ still builds under the current version of ghc and it provides support for generating Xilinx EDIF netlists that supports the basic FPGA circuit building blocks. If you need things like ODDRs etc. which appear in the new FPGAs then you have to either add them yourself (not too difficult) or wait for me to do it :) Cheers, Satnam ________________________________ Satnam Singh Microsoft 7 JJ Thomson Avenue Cambridge CB3 0FB United Kingdom Email: satnams(a)microsoft.com<mailto:satnams@microsoft.com> UK tel: +44 1223 479905 Fax: +44 1223 479 999 UK mobile: +44 7979 648412 USA cell: 206 330 1580 USA tel: 206 219 9024 URL: http://research.microsoft.com/~satnams Live Messenger: satnam(a)raintown.org<mailto:satnam@raintown.org>

1 0

Data.Derive problems
by Ron Alford 10 Jul '08

10 Jul '08

I'm using GHC 6.8.3 with $ cabal --version cabal-install version 0.5.1 using version 1.4.0.1 of the Cabal library I installed Data.Derive from hackage, only to be unable to find the 'derive' binary! Trying it directly from darcs, I get: $ ghc --make Setup.hs [1 of 1] Compiling Main ( Setup.hs, Setup.o ) Linking Setup ... silverback-wired:derive ronwalf$ ./Setup configure Warning: defaultUserHooks in Setup script is deprecated. Configuring derive-0.1.1... Warning: No 'build-type' specified. If you do not need a custom Setup.hs or ./configure script then use 'build-type: Simple'. -Ron

3 4

Poor Parsec error message
by Lyle Kopnicky 10 Jul '08

10 Jul '08

Hi folks, I'm using Parsec to parse a stream of tokens. The token primitive takes, among other arguments, a function to print tokens. However, this function is not always applied. Try the code below: --------- import Text.ParserCombinators.Parsec import Text.ParserCombinators.Parsec.Pos(newPos) mytoken :: (Eq t, Show t) => t -> GenParser (SourcePos,t) () t mytoken x = token showTok posFromTok testTok where showTok (pos,t) = "<" ++ show t ++ ">" posFromTok (pos,t) = pos testTok (pos,t) = if (x == t) then Just t else Nothing main = do putStrLn "" case parse the123Parser "" [(newPos "" 1 n, n) | n <- [1,2,3,4]] of (Left err) -> putStrLn (show err) (Right _) -> putStrLn "parsed correctly" putStrLn "" case parse the123Parser "" [(newPos "" 1 n, n) | n <- [1,3,4]] of (Left err) -> putStrLn (show err) (Right _) -> putStrLn "parsed correctly" the123Parser = do mytoken 1 mytoken 2 mytoken 3 eof return 123 ------- The output I get looks like this: (line 1, column 4): unexpected [((line 1, column 4),4)] expecting end of input (line 1, column 3): unexpected <3> In the second parse case, it correctly uses my showTok function to show the token. But in the first case, it just uses the regular show method. I guess that's because the eof parser doesn't know anything about how to show the token it sees. Any ideas on how I can get the error message in the first case to look more like the second case? Thanks, Lyle

1 1

Lazy IO
by Ronald Guida 10 Jul '08

10 Jul '08

Suppose I have a lazy function f :: [Int] -> [Int], and I happen to know that for all n, the n-th element of the output may only depend on the first (n-1) elements of the input. I want to print a number from f's output list, and then ask the user for the next number in f's input list, and then loop until the user stops providing valid numbers. I also need to be able to do IO after my loop exits. Consider the following code: ---------------------------------------------- module Main where import Control.Monad.Fix import System.IO.Unsafe promptInt :: String -> IO (Maybe Int) promptInt p = do putStr p s <- getLine let rs = reads s if not $ null rs then return $ Just $ fst $ head rs else return $ Nothing promptInts :: [String] -> IO [Int] promptInts [] = return [] promptInts (p:ps) = do m <- promptInt p case m of Just n -> do ns <- unsafeInterleaveIO $ promptInts ps return $ n:ns Nothing -> return [] -- assume accumulator is an opaque function accumulator :: [Int] -> [Int] accumulator = scanl (+) 0 makeAccPrompt :: Int -> String makeAccPrompt n = "[Acc = " ++ show n ++ "] ? " main :: IO () main = do xs <- mfix $ promptInts . map makeAccPrompt . accumulator seq (length xs) $ print xs ---------------------------------------------- Question: If I can't change my function f (in this case, accumulator), then is it possible to get the effect I want without having to resort to "unsafeInterleaveIO"?

3 2