September 2009 - Haskell-Cafe

Re: [Haskell-cafe] Essentials about type classes?
by Kim-Ee Yeoh 11 Sep '09

11 Sep '09

Hi Fredrik, Temaran wrote: > > data Dar = Dar String String > deriving (Show, Eq) > > class Bar a where > foo :: a -> Int > > instance Bar Dar where > foo(Dar n c) = length c > > but it keeps generating the same error; > > ERROR "./Bar.hs":16 - Inferred type is not general enough > *** Expression : foo > *** Expected type : a -> Int > *** Inferred type : Dar -> Int > You're missing indentation. Instead of declaring foo as a member of the typeclass, your code says it's a top-level function fully-polymorphic in the first parameter. That's why defining it later as foo (Dar ...) gives the error message it does. -- View this message in context: http://www.nabble.com/Essentials-about-type-classes--tp25393728p25405022.ht… Sent from the Haskell - Haskell-Cafe mailing list archive at Nabble.com.

1 0

Control.Exception base-3/base-4 woes
by Neil Brown 11 Sep '09

11 Sep '09

Hi, In my CHP library I need to do some exception catching. I want the library to work on GHC 6.8 (with base-3 -- this is the current version in Ubuntu Hardy and Jaunty, for example) and GHC 6.10 (which comes with base-4). But base-3 and base-4 need different code for exception catching (whether it's importing Control.OldException or giving a type to the catch method). Here's what I currently do -- my Haskell file contains this: #if __GLASGOW_HASKELL__ >= 609 import qualified Control.OldException as C #else import qualified Control.Exception as C #endif My cabal file contains this (it used to say just "base,..." but Hackage complained at me the other day when I tried to upload that): Build-Depends: base >= 3 && < 5, ... This works on two machines: one is 6.8+base-3, the other is 6.10+base-3&base-4, where cabal seems to use base-4. However, I have had a bug report (now replicated) which stems from a different 6.10+base-3&base-4 machine where cabal picks base-3 instead. The real problem is that the #if is based on GHC version, but really it should be based on which base-* library is being used. I know the code works with base-3 (use Control.Exception) and base-4 (use Control.OldException) but I can't get the build process right to alter the code based on which base-* library is being used. Can anyone tell me how to fix this? I don't think that changing to always use Control.Exception would fix this, because I need to give a different type for catch in base-3 to base-4, so there's still the incompatibility to be dealt with. Thanks, Neil.

5 4

Externally derive instance of Data?
by Dimitry Golubovsky 11 Sep '09

11 Sep '09

Hi, Given a datatype defined somewhere in a third-party package, without deriving (Data) specified. Is it possible, in my module which is importing that datatype from that package, to auto-derive instance of Data for the said datatype? The goal is not to recompile the package just because an auto-derived instance is needed. Or no way other than to recompile the package? Thanks. -- Dimitry Golubovsky Anywhere on the Web

2 1

Problem installing OpenGL
by Seb 11 Sep '09

11 Sep '09

Hi, I'm having trouble installing OpenGL through cabal. It fails when it tries to install OpenGLRaw with this error message: cabal: Missing dependency on a foreign library: * Missing C library: GL I'm running OS X Leopard and have Xcode 3.1.3 installed. Do i have to install something separate or is cabal just not looking in the right places?

1 1

Resolving overloading loops for circular constraint graph
by Stefan Holdermans 10 Sep '09

10 Sep '09

Manuel, Simon, I've spotted a hopefully small but for us quite annoying bug in GHC's type checker: it loops when overloading resolving involves a circular constraint graph containing type-family applications. The following program (also attached) demonstrates the problem: {-# LANGUAGE FlexibleContexts #-} {-# LANGUAGE TypeFamilies #-} type family F a :: * type instance F Int = (Int, ()) class C a instance C () instance (C (F a), C b) => C (a, b) f :: C (F a) => a -> Int f _ = 2 main :: IO () main = print (f (3 :: Int)) My guess is that the loop is caused by the constraint C (F Int) that arises from the use of f in main: C (F Int) = C (Int, ()) <= C (F Int) Indeed, overloading can be resolved successfully by "black-holing" the initial constraint, but it seems like the type checker refuses to do so. Can you confirm my findings? I'm not sure whether this is a known defect. If it isn't, I'd be more than happy to issue a report. Since this problem arises in a piece of very mission-critical code, I would be pleased to learn about any workarounds. Thanks in advance, Stefan

3 5

Fwd: Re[Haskell-cafe] [2]: memoization
by Alberto G. Corona 10 Sep '09

10 Sep '09

instead o that you can use a key such is: key :: a -> Int key = unsafePerformIO . hashStableName . makeStableName that is defined for any kind of data then, a unique key for the pair f x could be: key1 f x=(key f , key x) However my experience is that ocassionally gives different hashes for the same object, so maybe a few registers will be duplicated. 2009/9/10 <mf-hcafe-15c311f0c(a)etc-network.de> > On Thu, Sep 10, 2009 at 05:23:26AM -0700, staafmeister wrote: > > To: haskell-cafe(a)haskell.org > > From: staafmeister <g.c.stavenga(a)uu.nl> > > Date: Thu, 10 Sep 2009 05:23:26 -0700 (PDT) > > Subject: Re: Re[Haskell-cafe] [2]: memoization > > > > > > > > Hi Bulat, > > > > > > Bulat Ziganshin-2 wrote: > > > > > > Hello staafmeister, > > > > > > Thursday, September 10, 2009, 3:54:34 PM, you wrote: > > > > > >> What do you think about such a function? This function is > > > > > > a bit of refactoring > > > > > > -- "global variable" in haskell way > > > cache = unsafePerformIO $ newIORef M.empty > > > > > > memo f x = unsafePerformIO$ do > > > m <- readIORef cache > > > case M.lookup x m of > > > Just y -> return y > > > Nothing -> do let res = f x > > > writeIORef cache $ M.insert x > res m > > > return res > > > > > > memo2 = curry . memo . uncurry > > > > > > > This doesn't work and is exactly what I'm afraid the compiler is going to > > do. Cache needs to > > be associated with the function f. > > > > Otherwise one would get conflicts > > then make the cache object store functions together with values. > > > cache = unsafePerformIO $ newIORef M.empty > > memo f x = unsafePerformIO$ do > m <- readIORef cache > case M.lookup (mkKey f, x) m of > Just y -> return y > Nothing -> do let res = f x > writeIORef cache $ M.insert (mkKey > f, x) res m > return res > > memo2 = curry . memo . uncurry > > This leaves mkKey. Since functions are neither Ord nor Show, you'd > have to hack something together yourself. Perhaps an explicit > argument to memo? > > memo :: (Ord a) => String -> (a -> b) -> a -> IO b > memo fname f x = unsafePerformIO$ do > m <- readIORef cache > case M.lookup (fname, x) m of > Just y -> return y > Nothing -> do let res = f x > writeIORef cache $ M.insert (fname, > x) res m > return res > > there is probably a better and more elegant solution, but this should > at least work. right? > > > matthias > _______________________________________________ > Haskell-Cafe mailing list > Haskell-Cafe(a)haskell.org > http://www.haskell.org/mailman/listinfo/haskell-cafe >

1 1

Re: Snow Leopard breaks GHC
by Brian Sniffen 10 Sep '09

10 Sep '09

Having edited the Haskell Platform's /usr/bin/ghc in place, most packages install fine. I'm still having trouble with Pandoc, even given the advice: > Once cabal works, options --ld-option=-m32 (and also --gcc-option=-m32) > may be used. These options may also be passed to "./Setup configure" The problem appears to come when linking an incompatible zlib version: src/Text/Pandoc/ODT.hs:49:26: Exception when trying to run compile-time code: user error (Codec.Compression.Zlib: incompatible zlib version) Code: ($) makeZip "data" </> "odt-styles" In the first argument of `read', namely `$(makeZip $ "data" </> "odt-styles")' In the expression: read ($(makeZip $ "data" </> "odt-styles")) In the definition of `refArchive': refArchive = read ($(makeZip $ "data" </> "odt-styles")) The same problem occurs when making any call to Codec.Archive.Zip or Codec.Compression.Zlib. I do have a universal zlib installed by MacPorts, as well as the universal zlib that shipped with Snow Leopard and the universal zlib that came with Cabal. I'm not sure whether this message indicates that TH code is searching a different library path than non-TH code or what. Advice is most welcome. I'm particularly interested in finding out which zlib versions are being found at the construction of Codec.Compression.Zlib and at runtime (Pandoc compile time). -- Brian Sniffen http://evenmere.org/~bts/ <bts(a)evenmere.org>

3 7

retrieving arguments for functions from a heterogenous list (HList)
by Andrew U. Frank 10 Sep '09

10 Sep '09

I have a number of functions which have some arguments and produce a single result. all the arguments are in a heterogenous list and the results should update the list. this appears somewhat similar to the keyword-argument solution for functions proposed by oleg kiselyov in 2004. i would like to have a method to convert the functions f :: A -> B -> C g:: A -> D -> C -> B (for example) to f' :: argTypes -> H -> H g' :: argTypes -> H -> H where H is a heterogenous list containing A, B, C, D (and others) and argTypes is another heterogenous list containing the types of the arguments, which are used to retrieve the data from the list (with .!!.) - the result has the correct type and is updated in the list with .@@. . naively this could be done by some functins fx (recursing over the list of arguments) storing the last argument in the HList: fx :: (HList l) => op -> HCons e l -> H -> H fx op HNil uod = uod .@@. op fx op (HCons a as) uod = fx (op arg) as uod where arg = uod .!!. a this code does not compile, because HNil and HCons do not match (they are different data types, not a union type). code that compiles looks like class (HList l, HList k , HUpdateAtHNat n e k k, HType2HNat e k n , HLookupByHNat n k e ) => HCall e n f l k | l k -> n where hcall :: f -> l -> k -> k -- not general instance ( HNat n , HUpdateAtHNat n e UoD UoD, HType2HNat e UoD n , HLookupByHNat n UoD e ) => HCall e n e HNil UoD where hcall opv HNil k = hcall0 opv k instance (HCall e n f l UoD , HLookupByHNat n UoD e, HType2HNat e n UoD ) => HCall e n (e -> f) (HCons a l) UoD where hcall op (HCons a l) k = hcall (op arg) l k where arg = hLookupByHNat n k n = hType2HNat (toProxy arg) k :: n hcall0 :: ( HUpdateAtHNat n f UoD UoD, HType2HNat f H n, HCall e n f HNil UoD) => f -> H -> H hcall0 op k = hUpdateAtHNat n (const op undefined) k `asTypeOf` k where n = hType2HNat (toProxy op) k where i factored out the terminating call, which is probably wrong. i would like to use it as: showPop4 :: (HList hNil, HCall Population n (District -> Population) (hConst District hNil) UoD) => (hCons District hNil) -> UoD -> UoD showPop4 arg uod = hcall showPop2 arg uod unfortunately, calls to this function result in 'cannot deduce' compilation error; it seems that some of the intermediate types are not known. i appreciate any help! andrew

1 0

Dutch HUG: meeting tomorrow in Utrecht
by Tom Lokhorst 10 Sep '09

10 Sep '09

Hi everyone, I want to remind all functional programmers currently in The Netherlands of the Dutch Haskell User Group [1, 2] meeting tomorrow at 19:00 [3]. The meeting will be in Booth Hall [4] of the Utrecht University Library, where we'll have three talks. There's free car parking [5] at the library, and it's easily accessible by bike and bus [6]. Also, to stay on top of Dutch HUG related news, please subscribe to the mailing list [2]. See you tomorrow! - Tom Lokhorst [1]: http://www.haskell.org/haskellwiki/Dutch_HUG [2]: http://groups.google.com/group/dutch-hug [3]: http://www.mail-archive.com/haskell-cafe@haskell.org/msg64036.html [4]: http://www.uu.nl/EN/library/contact/university_library/zaalverhuur/Pages/de… [5]: http://www.uu.nl/EN/library/contact/university_library/Parkeren/Pages/defau… [6]: http://www.uu.nl/EN/library/contact/university_library/plattegrondenrou/Pag…

1 0

Would you mind explain such a code ?
by zaxis 10 Sep '09

10 Sep '09

myFoldl :: (a -> b -> a) -> a -> [b] -> a myFoldl f z xs = foldr step id xs z where step x g a = g (f a x) I know myFoldl implements foldl using foldr. However i really donot know how it can do it ? Please shed a light one me, thanks! -- View this message in context: http://www.nabble.com/Would-you-mind-explain-such-a-code---tp25377949p25377… Sent from the Haskell - Haskell-Cafe mailing list archive at Nabble.com.

5 6