April 2011 - Beginners

newbie seeking code review
by Sean Perry 28 Apr '11

28 Apr '11

I just finished "Learn you a Haskell". Fine book, really enjoyed it. Early on he develops a very simple RPN evaluator. I played with it a bit and expanded it to use Either instead of Maybe. Also, '_' recalls the last value so you can do: > 10 5 + 15.0 > _ 3 * 45.0 > _ 5 / 9.0 This is one of my first complete pieces of Haskell. I would appreciate comments on style, formatting, etc. There are comments in places I am seeking particular guidance. If you are also reading along in LYAH, warning spoiler alert...... import Control.Monad (foldM, liftM) import Control.Monad.Error import System.Exit (exitFailure, exitSuccess) import System.IO (hFlush, stdout) import System.IO.Error (isEOFError) readNumMaybe :: String -> Either String Double readNumMaybe st = case reads st of [(x, "")] -> Right x [(x, s)] -> Left $ "incomplete parse: '" ++ st ++ "'" [] -> Left $ "unknown value: '" ++ st ++ "'" evalRPN :: [Double] -> String -> Either String [Double] evalRPN _ "clear" = return [] evalRPN (x:y:xs) "+" = return $ (y + x):xs evalRPN (x:y:xs) "-" = return $ (y - x):xs evalRPN (x:y:xs) "*" = return $ (y * x):xs evalRPN (x:y:xs) "/" = return $ (y / x):xs evalRPN (x:y:xs) "^" = return $ (y ** x):xs evalRPN (x:xs) "ln" = return $ (log x):xs evalRPN xs "sum" = return $ [sum xs] evalRPN xs num = liftM (:xs) (readNumMaybe num) evalRPNTokens :: [String] -> Either ([String], String) (Maybe Double) evalRPNTokens tokens = case foldM evalRPN [] tokens of Right [] -> Right Nothing Right [x] -> Right (Just x) Right result -> Left $ (tokens, show (length tokens) ++ " items left on stack.") Left msg -> Left (tokens, msg) -- is this used with map ok or am I missing a standard function? replaceif p new s = if s == p then new else s showEval :: String -> Either ([String], String) (Maybe Double) -> IO String showEval _ (Right Nothing) = return "" showEval _ (Right (Just result)) = do -- show result then use this value for the next iteration print result return $ show result showEval last (Left (input, msg)) = do putStrLn $ "Error: '" ++ unwords input ++ "': " ++ msg -- do not let errors eat the last successful value return last getLineOrQuit = catch getLine (\e -> if isEOFError e then putStr "\n" >> exitSuccess else exitFailure) prompt = "> " evalLoop :: String -> IO () evalLoop last = do putStr prompt hFlush stdout -- ensure the prompt is shown -- is this too "cute"?? result <- getLineOrQuit >>= showEval last . evalRPNTokens . map (replaceif "_" last) . words evalLoop result main = evalLoop ""

2 1

Understanding program stack usage
by Patrick LeBoutillier 28 Apr '11

28 Apr '11

Hi, I've written a program that reads doubles from stdin and creates a distribution of the data according to specified ranges. The program works ok and speed is acceptable (17 secs for 300000 entries), but to run it I have to increase stack size: $ cat input | ./distrib +RTS -K100000000 -RTS How do I figure out why the program needs a lot of stack? Also, is there an easy way to increace performance? Thanks, Patrick import qualified Data.Map as M import Data.List (sort, foldl') import qualified Data.ByteString.Char8 as L import Data.ByteString.Lex.Double data Range n = Range n n | OutOfBounds deriving (Eq, Ord, Show) data Distrib n = Distrib (M.Map (Range n) Int) [Range n] deriving (Show) mkDist :: (Ord n) => [n] -> Distrib n mkDist ns = Distrib M.empty ranges where ranges = zipWith Range l (tail l) l = sort ns record :: (Ord n) => n -> Distrib n -> Distrib n record n (Distrib m rs) = Distrib (M.alter f slot m) rs where f (Just n) = Just $ n + 1 f Nothing = Just 1 slot = findSlot n rs findSlot x (r@(Range a b):rs) | x >= a && x < b = r | otherwise = findSlot x rs findSlot x [] = OutOfBounds mkDouble :: L.ByteString -> Double mkDouble bs = case readDouble bs of Just (d, rest) -> d main = do let d = mkDist [0, 5, 10, 50, 100] bs <- L.getContents let ds = map (abs . mkDouble) . L.lines $ bs let (Distrib m _) = foldr (\n acc -> record n acc) d ds putStrLn . show $ m -- ===================== Patrick LeBoutillier Rosemère, Québec, Canada

4 6

Good reads?
by Christopher Svanefalk 27 Apr '11

27 Apr '11

Hey all, I have mostly been brought in imperative/OOP languages. In connection with my college studies, however, I have taken great interest in the functional paradigm, and would very much like to dig deeper into it. I am currently reading through Peyton-Jones "Haskell: The Craft of Functional Programming" (2nd ed.), as well as a great paper published by one of my professors (http://www.cse.chalmers.se/~rjmh/Papers/whyfp.html) However, what other works, in your opinions, should I look into to get a more complete understanding of functional programming? Thanks in advance! Chris

1 0

partition a list in all possible ways
by I. J. Kennedy 27 Apr '11

27 Apr '11

I'm looking for a function [a] -> [[[a]]] that will partition a list into non-empty pieces in all possible ways. For example f [1,2,3,4] = [[[1],[2],[3],[4]], [[1],[2],[3,4]], [[1],[2,3],[4]], [[1],[2,3,4]], [[1,2],[3],[4]], [[1,2],[3,4]], [[1,2,3],[4]], [[1,2,3,4]]] Perhaps this is a well-known function to experts, but not to me. Hoogle doesn't seem to have anything with this signature. Thanks, Jack

3 2

Copy record with mutations?
by Andrew n marshall 24 Apr '11

24 Apr '11

Is there a syntax or function to copy a record, but with select fields populated with new values? For example, extending LYAH's Car let stang67 = Car {company="Ford", model="Mustang", year=1967} let stang68 = stang67 `mutate` Car {year=1968} such that company and model are "inherited".

4 5

install from Hachage
by Luca Ciciriello 22 Apr '11

22 Apr '11

Hi All. I have to programmatically build and install a HachageDB package. I've found some examples using "Distribution.Simple.SetupWrapper" but this module is no longer available in Cabal 1.10.x Is there something equivalent I can use? Thanks in advance for any answer. Luca.

1 0

dealing with chained functions
by Alec Benzer 21 Apr '11

21 Apr '11

Is there a generally accepted "better" way of dealing with the following: say I have three functions that all rely on output from each other. Ie, we get raw input, which one function processes. another function then requires the output from this function and produces some other output, which a third function then requires. would it be better to implement the functions like this: f1 :: a -> b f2 :: b -> c f3 :: c -> d final :: a -> d or like this: f1 :: a -> b f2 :: a -> c f3 :: a -> d Ie, in the latter set, f2 calls f1 on its input, and then proceeds to process it. f3 calls f2 on its input, etc. In the former case, final would just chain all three functions together itself (or I might just do it in main or somewhere else that's appropriate) Are one of these methods generally considered a better practice than the other, specifically in a functional context? My gut tells me that the former method looks more appropriate, but I'm not sure if it'd really be the accepted practice regardless of context, or if one might use either of these methods depending on the specifics of what they're doing. Or if it's just a matter of taste/opinion?

4 6

(no subject)
by Luca Ciciriello 20 Apr '11

20 Apr '11

Hi All.I'm using GHC with MacOS X 10.6.7 (Xcode4). I've installed GHC 7.0.3 and the HackageDB package hsgsom. Then, for my motivation, I've uninstalled GHC.To remove GHC I've used the uninstaller tool and I've manually removed the folder /Library/Frameworks/GHC.framework. I've also manually removed the folder /Users/MyUser/Library/Haskell and the folder /Users/MyUser/.cabalNow I've installed GHC again, but when I try to install the package hsgsom cabal tells to me that the package is already installed and I have to use the --reinstall flag. So, where the information of the installed packages are stored on my system? How can I remove all Haskell dependencies from my system in order to start with a clean installation?Thanks in advance for any answer.Luca.

3 5

linker Warning
by Luca Ciciriello 20 Apr '11

20 Apr '11

Hi All.I've installed on my MacOS X 10.6.7 (Xcode4) the new GHC 7.0.3 (64 bit)All works fine, but every time I build my code I get the warning:ld: warning: could not create compact unwind for _ffi_call_unix64: does not use RBP or RSP based frameIs this warning something bad?Thanks.Luca.

1 0

Hackage documentation generation
by Patrick LeBoutillier 20 Apr '11

20 Apr '11

Hi, How long does it take for the haddock docs to be generated on Hackage? I uploaded a package 2 days ago and the docs are not there. Is there a problem with my package? Patrick -- ===================== Patrick LeBoutillier Rosemère, Québec, Canada

3 2