August 2012 - Beginners

complex typeclass/constraint question
by Dennis Raddle 30 Aug '12

30 Aug '12

{- I'm writing code to do some combinatorial construction via test-and-evaluate algorithms. And I want to write a very generic algorithm. The idea is that the item being constructed has a current state, which is represented by a data type of typeclass EvalState. And the algorithm will consider a set of new elements to add to the object being constructed, represented by a data type of typeclass ElemSet. Here are the class definitions: -} class ElemSet a where elemSetNumElements :: a -> Int class EvalState s where isCompleteState :: s -> Bool newElemSet :: ElemSet a => s -> a integrateElem :: ElemSet a => a -> Int -> s -> s -- given an elem set, and an index of the elem to choose , compute a score scoreElem :: ElemSet a => a -> Int -> s -> EvalScore type EvalScore = Float type RandomnessChooser = [(EvalScore,Int)] -> ErrorRand Int {- here's my generic search algorithm written using the typeclasses the problem I'm having relates to the fact that I need to call 'integrateElem' which is a function in typeclass EvalState, and one of the arguments has the constraint of being of typeclass ElemSet, but I don't know how to put that constraint in. The specific error is "Ambigous type variable 'a' in the constraint: 'ElemSet a' arising from a use of 'integrateElem'" -} search :: (EvalState s) => RandomnessChooser -> s -> ErrorRand ssearch chooser state = do if isCompleteState state then return state else do let elemSet = newElemSet state n = elemSetNumElements elemSet scores = map g [0..n-1] g i = (scoreElem elemSet i state,i) chosenElemIdx <- chooser scores let newState = integrateElem elemSet chosenElemIdx state search chooser newState

2 2

How to load multiple source files in GHCi?
by Yang 30 Aug '12

30 Aug '12

Hi, I have a few source files in current directory to load as modules of my library in GHCi. How should I do with :load command? I tried ghci>:load a.hs where module a is defined and exported for testing purpose, but failed for a.hs:1:1: File name does not match module name: Saw: `Main' Expected: `a' Failed, modules loaded: none. However, I do not need Main to run for testing. I just want to test exported functions. -- Best Regards Yang Zhao

2 1

IO vars
by Corentin Dupont 29 Aug '12

29 Aug '12

Hi all, there is something very basic that it seems escaped me. For example with the following program f and g have type IO () and I can thread a value between the two using a file. Can I do the exact same (not changing the types of f and g) without a file?* f,g :: IO () **f = withFile "toto" WriteMode (flip hPutStr "toto") g = withFile "toto" ReadMode hGetLine >>= putStrLn main = f >> g* Thanks and cheers, Corentin

5 6

Re: [Haskell-beginners] IO vars & ACID
by Corentin Dupont 29 Aug '12

29 Aug '12

To give you more context, I have a soft using ACID state to store and retrieve values. For example I have a function like that: newPlayer :: PlayerName -> IO () newPlayer name = update $ AddPlayer name It seems that ACID uses a global state or a file to store the events... I'd like to suppress the ACID state for the moment, how should I do? Change all the* IO () *types to *StateT Game IO ()*? Indead the ACID state is started with: c <- localStartSystemState (Proxy :: Proxy Game) Best, C On Wed, Aug 29, 2012 at 12:00 PM, Corentin Dupont <corentin.dupont(a)gmail.com > wrote: > Thanks Eugene and Ozgur. > I also looked on the side of IORef, but it doesn't looked to be a good > solution since we have to use unsafePerformIO. > > I have a big program to modify, and I want to pass some new data to > existing functions of type IO(). I'd like to avoid changing all the > function's types down the chain... What is the best way to do that? > > > > > On Wed, Aug 29, 2012 at 11:43 AM, Ozgur Akgun <ozgurakgun(a)gmail.com>wrote: > >> On 29 August 2012 10:21, Corentin Dupont <corentin.dupont(a)gmail.com>wrote: >> >>> *f,g :: IO () >>> f = withFile "toto" WriteMode (flip hPutStr "42") >>> g = withFile "toto" ReadMode hGetLine >>= (\s -> putStrLn $ "Answer:" ++ >>> s) >>> main = f >> g* >>> >>> Is it possible to do the same without files (the types must remain IO())? >>> >> >> One can use an IORef to get a similar effect. >> >> import Data.IORef >> import System.IO.Unsafe >> >> {-# NOINLINE toto #-} >> toto :: IORef String >> toto = unsafePerformIO (newIORef "") >> >> f,g :: IO () >> f = writeIORef toto "42" >> g = readIORef toto >>= (\s -> putStrLn $ "Answer:" ++ s) >> >> main = f >> g >> >> HTH, >> Ozgur >> > >

1 0

Are tuples really needed?
by Carlos J. G. Duarte 29 Aug '12

29 Aug '12

Sorry if this question is too insane, but I was wondering if tuples are really needed in Haskell. I mean, could tuples be generally replaced by variables unroll (f x y z) and explicit data types, or are there some things only possible to do via tuples? Thx in advance (and sorry if this looks silly).

5 5

"system" call uses a different shell, or does not pick up the whole environment
by Hong Yang 29 Aug '12

29 Aug '12

Hi, I am trying to mimic mapM() at shell command line. I define the interface as "mapm cmd2 cmd1," so cmd2 will be run for each of the cmd1 results. "$_" can be used inside cmd2 to represent the current cmd1 result. For example, the command mapm 'cp -pr $_ destination_dir/$_' ls copies everything under the current directory to the destination directory. The code is as follows: -- module Main where import System.Environment ( getArgs ) import System.Exit import System.IO import System.Process import Text.Regex import Text.Regex.Posix main = do hs_argv <- getArgs if length hs_argv /= 2 then putStrLn "wrong arguments!" >> exitFailure else do let [cmd2, cmd1] = hs_argv (_, hOut, hErr, _) <- runInteractiveCommand cmd1 err <- hGetContents hErr hClose hErr if null err then do out <- hGetContents hOut mapM (f cmd2) (lines out) else putStr err >> exitFailure f :: String -> String -> IO ExitCode f cmd2 item = system cmd2' where cmd2' = if cmd2 =~ "\\$\\_"::Bool then subRegex (mkRegex "\\$\\_") cmd2 item else cmd2 -- It works, except one issue that is bothering me. If I issue mapm 'lt $_' ls, I get a bunch of /bin/sh: lt: command not found, while I expect it act the same as mapm 'ls -Alrt --color=auto $_' ls, because "lt" is aliased to "ls -Alrt --color=auto." Notice "/bin/sh" above. My shell is actually tcsh. All the aliases are in ~/.cshrc. I tried replacing "system cmd2'" with system ("source ~/.cshrc; " ++ cmd2') and system ("tcsh -c " ++ "'source ~/.cshrc; " ++ cmd2' ++ "'"), but they did not solve the problem. Can someone please help me? Thanks, Hong

5 7

Errors involving rigid skolem types
by Matthew Moppett 28 Aug '12

28 Aug '12

The following code is intended as a first step towards creating a cyclical enumerable type, such that: (e.g.) [Cyc Friday .. Cyc Tuesday] would yield [Friday, Saturday, Sunday, Monday, Tuesday] module Cycle where newtype Cyc a = Cyc a deriving (Eq, Ord, Bounded, Show, Read) fromCyc :: Cyc a -> a fromCyc (Cyc a) = a instance (Enum a, Bounded a) => Enum (Cyc a) where fromEnum = fromEnum . fromCyc toEnum n = Cyc x where (x, max) = (x', maxBound) :: (a, a) x' = toEnum $ n `mod` ((fromEnum max) - 1) This yields a kind of error message that I've often bashed my head against in other code I've written, without ever really understanding what the problem is exactly: Couldn't match type `a0' with `a1' because type variable `a1' would escape its scope This (rigid, skolem) type variable is bound by an expression type signature: (a1, a1) The following variables have types that mention a0 x' :: a0 (bound at Cycle.hs:12:15) In the expression: (x', maxBound) :: (a, a) In a pattern binding: (x, max) = (x', maxBound) :: (a, a) In an equation for `toEnum': toEnum n = Cyc x where (x, max) = (x', maxBound) :: (a, a) x' = toEnum $ n `mod` ((fromEnum max) - 1) The problem comes up when I'm trying to give hints to the compiler about the type that a particular expression should have. My questions are: (1) what exactly is going on here, and (2) is there any general technique for specifying types in situations like this that gets around this problem?

2 2

IO / State / Do Syntax
by Christopher Howard 27 Aug '12

27 Aug '12

So, I'm still working with my resource tracker idea -- keeping images stored and organized inside a resource tracker structure that I can pass around to functions that need them. Let's say I want a function like so: code: -------- initResources :: IO ResourceTracker -------- The idea being that initResources loads the image files, stores them in the resource tracker (RT) structure, and returns the RT. I know I can do something like so: code: -------- initResources = do pic1 <- loadImage "someimage.png" -- IO function pic2 <- loadImage "someimage2.png" -- ... and so on ... let rt = emptyResourceTracker in let rt' = storeImage rt "pic1keyword" pic1 in let rt'' = storeImage rt' "pic2keyword" pic2 in -- ... and so on, until finally: ... rt'''''''''''''''''''''''''' -------- Obviously, all the let statements and apostrophes are undesirable. So, presumably what I need is to being using the State monad, yes? (I must confess I have only a vague understanding of the State monad, even after reading several tutorials.) But in my initResources function, how do I mix use of the IO and State do syntax, and still get what I want? I think this has something to do with Monad transformers, but I'm even less clear on how those work. -- frigidcode.com indicium.us

2 1

How to switch off a RadioMenuiItem?
by Pierre Michallet 27 Aug '12

27 Aug '12

Using Glade, I have created a menu which,among other things,has 2 radioMenuItems located in the same submenu. Both are in the off state when the program starts. In the following, I will refer to them as "initiationCB" and "beginnerCB". The expected behavior is that these 2 options are exclusive ie clicking one when in the OFF state will turn it ON and turn the other OFF. In fact, the first time I click on "initiationCB" its associated checkbox is turned ON; a second click on it will not change its state. When "beginnerCB" is then clicked, it is turned ON but "initiationCB" is not turned OFF as expected, and from then on, the two items are ON and there is no way to turn them OFF. One surprising thing is that in the same menu there are 2 CheckMenuItems which, when clicked, change their state without any code written to obtain this behavior. I would have expected the same from RadioMenuItems. In an attempt to switch a RadioMenuItem OFF , I have added the following code: on initiationCB menuItemActivate do checkMenuItemSetActive initiationCB True When "initiationCB" is clicked (and turned ON as a consequence) the program blows up seeming in an infinite recursion. The Gtk2HS tutorial (chapter 7.2) gives an example of selective choices by defining actions in the file called by a UIManager to get the elements of the user interface and RadioActionEntry in the code. The problem is that I could not succeed in associating an action to a menu item in the Glade file (as done in the tutorial), and in case of success, I don't know whether a XmlGetWidget instruction would recognize it. So the question is : is there a way to implement a solution similar to the one described in the tutorial using Glade instead of the UIManager approach? Another odd thing to me is that the signal menuItemToggle is not raised when a radioMenuItem is clicked. Thanks in advance for your help p.michallet(a)free.fr

1 0

MonadError
by Dennis Raddle 26 Aug '12

26 Aug '12

I'm enjoying writing code that uses the Error monad to provide errors, which I can catch and rethrow to add context. Kind of like getting a stack trace with any context I want every time an error happens. In most cases I use monad transformers, so the actual monad type would be something like "ErrorT String (State Int) a" But much of my code that handles errors doesn't need to know the entire context. It uses throwError, but otherwise doesn't care anything else about the type of the monadic computation. I.e., it needs to "know" that it's inside a monadic computation of a typeclass MonadError. So I discovered that I could declare a function like this: x :: MonadError String m => Float -> m Float x y = if y == 0 then throwError "zero" else return y My awareness of class constraints is vague, so as I worked on this problem I knew in a fuzzy way it was going to look something like that, but I had to experiment. Also, because I put "String" in the class constraint, I need to turn on FlexibleContexts. So my question is, how would I do it without making "String" explicit? If I put x:: MonadError e m => Float -> m Float x y = if y == 0 then throwError "zero" else return y I get errors saying "couldn't match expected type e against inferred type [Char]" and something about functional dependencies I fiddled with it a bit but couldn't make it work.

2 1