June 2009 - Haskell-Cafe

Take a look at my photos on Facebook
by Andrew Smith 08 Jun '09

08 Jun '09

Hi haskell-cafe(a)haskell.org, I set up a Facebook Profile where I can post my pictures, videos and events and I want to add you as a friend so you can see it. First, you need to join Facebook! Once you join, you can also create your own profile. Thanks, Andrew To join Facebook, please follow the link below: http://www.facebook.com/p.php?i=675326041&k=RZM5Y4QZR5WMZAEIS15UY&r haskell-cafe(a)haskell.org was invited to join Facebook by Andrew Smith. If you do not wish to receive this type of email from Facebook in the future, please click on the link below to unsubscribe. http://www.facebook.com/o.php?k=917e28&u=531922198&mid=967b97G1fb47d16G0G8 Facebook's offices are located at 1601 S. California Ave., Palo Alto, CA 94304.

2 1

ghci: can't load .so/.DLL for: m (addDLL: could not load DLL) / is there a libm.dll for Windows
by Mark Wassell 08 Jun '09

08 Jun '09

Hello, I get this when using the logfloat package via ghci on Windows. For example *Main> :m + Data.Number.LogFloat *Main Data.Number.LogFloat> logFloat (1.0::Float) Loading package syb ... linking ... done. Loading package array-0.2.0.0 ... linking ... done. Loading package logfloat-0.12.0.1 ... can't load .so/.DLL for: m (addDLL: could not load DLL) *Main Data.Number.LogFloat> When compiling the program using ghc, it works fine. This was previously reported as a ghc bug http://hackage.haskell.org/trac/ghc/ticket/3242 where the reporter was getting the message using hipmunk. The ticket was closed as the issue was deemed to be a bug in hipmunk, not GHC. As the ticket reporter asked, what is the correct approch to get around this? I have tried specifying -lm on ghci startup line but my system doesn't seem to have a libm.dll or anything similar. Is there such a thing? Thanks Mark

2 1

Roman to Decimal Algorithms
by Andrew Savige 07 Jun '09

07 Jun '09

I recently played in a code golf Roman to Decimal challenge (in Perl, Python, Ruby and PHP). In playing this game, I found some interesting short algorithms for converting from Roman Numerals to Decimal. Because I think some of these are new, I'd like to present them here, in case they are not really new after all, or in case there are problems with these algorithms that I've overlooked. I'd like to eventually write up a "rosetta code" article comparing the implementation of these algorithms in various languages, including Haskell. Since I'm only a very occasional Haskell programmer, I thought it best to get feedback from Haskell experts before inflicting any of my Haskell code on a wider audience. Hence this post. To keep this post reasonably brief, note that these algorithms are for "modern" Roman Numerals only, limited to the range 1-3999, and with no error checking. That is, the code below assumes that the input is always a well formed Roman Numeral. I have tested each algorithm against every modern Roman Numeral in the range 1-3999. As a starting point, note this "HaskellWiki" function: romanToInt :: String -> Int romanToInt = fst . foldr (\p (t,s) -> if p >= s then (t+p,p) else (t-p,p)) (0,0) . map (fromJust . flip lookup (zip "IVXLCDM" [1,5,10,50,100,500,1000])) taken from http://haskell.cs.yale.edu/haskellwiki/Roman_numerals. I'm going to essentially duplicate the functionality of this code. {-# OPTIONS_GHC -fglasgow-exts -Wall #-} import Data.Char (toUpper) rtoa :: Char -> Int rtoa 'M' = 1000 rtoa 'D' = 500 rtoa 'C' = 100 rtoa 'L' = 50 rtoa 'X' = 10 rtoa 'V' = 5 rtoa 'I' = 1 rtoa r = error $ "Invalid rtoa char:" ++ show r urtoa :: Char -> Int urtoa = rtoa . toUpper romanToInt :: String -> Int romanToInt = foldl1 (\t n -> t+n-t`mod`n*2) . map urtoa The essential difference between this solution and the HaskellWiki one is the use of the "running total" for state rather than the "previous value". I see this as an improvement mainly because it is shorter -- though you might argue that it is less clear. An alternative way to express the romanToInt function is: romanToInt = foldl (\t c -> t+(urtoa c)-t`mod`(urtoa c)*2) 0 I'm open to persuasion as to which is better Haskell style. Since I'm not playing golf anymore, the rtoa function above, though hardly short, seemed to me to be the simplest and clearest way to express converting a single Roman Numeral to its corresponding arabic number. Again, suggestions for the "best"/most efficient way to do this in Haskell are most welcome. A common, and often winning technique, in golf is to perform these sorts of conversions by concocting a "magic formula". For fun, I rewrote rtoa using a magic formula I used in the golf game: rtoa c = 10^(205558`mod`(ord c)`mod`7)`mod`9995 I'm not suggesting that magic formulae are useful outside of golf and this second rtoa function, though shorter, is much less clear. I might add that this particular magic formula appears to be less useful in Haskell golf than the other languages because `mod` is five times longer than the % operator of the other languages. :) By way of explanation, notice that this formula: 205558`mod`(ord c)`mod`7 maps I->0, X->1, C->2, M->3, V->4, L->5, D->6 as shown below: Roman m 10^m 10^m`mod`9995 ----- - ----- ------------- M 3 1000 1000 D 6 1000000 500 C 2 100 100 L 5 100000 50 X 1 10 10 V 4 10000 5 I 0 1 1 Noticing this, you can replace the 205558`mod`(ord c)`mod`7 magic formula with a function that returns a string index (index() in Perl and Python). I am sometimes overwhelmed by the quantity and richness of all the functions in the GHC Haskell libraries. I eventually found a Haskell solution that seemed to work: {-# OPTIONS_GHC -fglasgow-exts -XOverloadedStrings -Wall #-} import Data.ByteString.Char8 (elemIndex) import Data.Maybe (fromJust) rtoa c = 10^(fromJust (elemIndex c "IXCMVLD"))`mod`9995 I got this to work by trial and error and have no clue what this "-XOverloadedStrings" and "Data.ByteString.Char8" business really means. If there is a better Haskell way of finding the numeric index of a particular character in a string, please let me know. Alternatively, you could write the rtoa function using an approach taken from the original HaskellWiki solution: rtoa = fromJust . flip lookup (zip "IVXLCDM" [1,5,10,50,100,500,1000]) What is your recommendation as to the "best"/most efficient way of writing the rtoa function in Haskell? Finally, here is an example complete test program. Suggestions for improving the style of this code are welcome. {-# OPTIONS_GHC -fglasgow-exts -Wall #-} import Data.Char (toUpper) import Data.List (concat, intersperse) rtoa :: Char -> Int rtoa 'M' = 1000 rtoa 'D' = 500 rtoa 'C' = 100 rtoa 'L' = 50 rtoa 'X' = 10 rtoa 'V' = 5 rtoa 'I' = 1 rtoa r = error $ "Invalid rtoa char:" ++ show r urtoa :: Char -> Int urtoa = rtoa . toUpper -- one is derived from http://haskell.cs.yale.edu/haskellwiki/Roman_numerals one :: String -> Int one = fst . foldr (\p (t,s) -> if p >= s then (t+p,p) else (t-p,p)) (0,0) . map urtoa two :: String -> Int two = foldl1 (\t n -> t+n-t`mod`n*2) . map urtoa three :: String -> Int three = foldl (\t c -> t+(urtoa c)-t`mod`(urtoa c)*2) 0 myshow :: String -> (String -> Int) -> String -> String myshow capt fn val = capt ++ val ++ " " ++ (show (fn val)) testdata :: [String] testdata = [ "I", "xiv", "DXVI", "CMLXIII" ] main :: IO () main = do putStrLn $ (concat $ intersperse "\n" (map (\c -> (myshow "one : " one c)) testdata)) putStrLn $ (concat $ intersperse "\n" (map (\c -> (myshow "two : " two c)) testdata)) putStrLn $ (concat $ intersperse "\n" (map (\c -> (myshow "three: " three c)) testdata)) Cheers, /-\ Need a Holiday? Win a $10,000 Holiday of your choice. Enter now.http://us.lrd.yahoo.com/_ylc=X3oDMTJxN2x2ZmNpBF9zAzIwMjM2MTY2MTMEdG1fZG…

3 4

Re: [Haskell-cafe] Random Number
by michael rice 07 Jun '09

07 Jun '09

Good essay. Try this one for a laugh: http://www.mcs.vuw.ac.nz/comp/Publications/CS-TR-02-9.abs.html A good place to begin is PDF pg. 19. Michael --- On Sun, 6/7/09, Krzysztof Skrzętnicki <gtener(a)gmail.com> wrote: From: Krzysztof Skrzętnicki <gtener(a)gmail.com> Subject: Re: [Haskell-cafe] Random Number To: "ptrash" <ptrash(a)web.de> Cc: haskell-cafe(a)haskell.org Date: Sunday, June 7, 2009, 3:55 PM On Sun, Jun 7, 2009 at 21:33, ptrash<ptrash(a)web.de> wrote: > > Hi, > > is the are way (or a build in method) in haskell to get a random number from > a number bottom to a number top? > > Something like > > let randomNumber = random 1 30 > > to get a random number between 1 and 30. I don't mean to be rude, but did you even tried to read the documentation? The function you want is here: http://www.haskell.org/ghc/docs/latest/html/libraries/random/System-Random.… Before you ask any other questions please read this essay: http://mattgemmell.com/2008/12/08/what-have-you-tried Best regards Krzysztof Skrzętnicki _______________________________________________ Haskell-Cafe mailing list Haskell-Cafe(a)haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe

1 0

Random Number
by ptrash 07 Jun '09

07 Jun '09

Hi, is the are way (or a build in method) in haskell to get a random number from a number bottom to a number top? Something like let randomNumber = random 1 30 to get a random number between 1 and 30. -- View this message in context: http://www.nabble.com/Random-Number-tp23914474p23914474.html Sent from the Haskell - Haskell-Cafe mailing list archive at Nabble.com.

4 3

Change value of a variable
by ptrash 07 Jun '09

07 Jun '09

Hi, how can I change the value of a variable. let x = 1 x = x + 2 First I set the value of x to 1. Then I want to increase it by 2. This way doesn't work, because I think it is a infinite expression. Is there a way to change the value? -- View this message in context: http://www.nabble.com/Change-value-of-a-variable-tp23913404p23913404.html Sent from the Haskell - Haskell-Cafe mailing list archive at Nabble.com.

6 9

Question on rank-N polymorphism
by Vladimir Reshetnikov 07 Jun '09

07 Jun '09

Hi, I have the following code: -------------------------------------------------------------------------------- fs g = (g fst, g snd) examples = (fs fmap, fs liftA, fs liftM, fs id, fs ($(1,"2")), fs ((,)id), fs (:[]), fs repeat) -------------------------------------------------------------------------------- The idea is that fs accepts a polymorphic function as its argument. What type signature can I specify for f in order to compile this code? If it is not possible in Haskell, is there another language with static typing which allows this? Thanks, Vladimir

3 3

comprehension problem
by ptrash 07 Jun '09

07 Jun '09

Hi, what does this to code rows mean: newtype Probability = P Float newtype Dist a = D {unD :: [(a, Probability)]} newtype definies a new type called Probability. But what does P Float mean? And what is the a in Dist a? What does D {...} mean? Thanks for your help. -- View this message in context: http://www.nabble.com/comprehension-problem-tp23858359p23858359.html Sent from the Haskell - Haskell-Cafe mailing list archive at Nabble.com.

2 2

totally confused about Haskell namespace issue conventions
by Vasili I. Galchin 07 Jun '09

07 Jun '09

Hello, Should namespace designation be specified in modules or in the .cabal file? Or to put it another way should a relative namespace be specified in a Haskell module and the remaining "top" part be specified in the associated .cabal file? Of course, yes/no answers are probably not sufficient ... i.e. please elaborate. Kind regards, Vasili

3 4

Re: [Haskell-cafe] Question on rank-N polymorphism
by Ryan Ingram 07 Jun '09

07 Jun '09

Well, I don't really recommend programming in dependently typed languages right now :) But if you must, Agda has been getting a lot of attention recently. Also, the theorem prover Coq is based on the dependently-typed lambda calculus. In Haskell, giving a function an intersection type is generally done with typeclasses. You can write, for example: class Fs a where type FsResult a fs :: a -> FsResult a data Fmap = Fmap instance Fs Fmap where type FsResult a = forall f a b. Functor f => (f (a,b) -> f a, f (a,b) -> f b) fs Fmap = (fmap fst, fmap snd) (although this seems unusable to me!) You can also use Template Haskell to copy the argument: -- I may have the syntax wrong here fs :: a -> Q Exp fs a = [e| ($a fst, $a snd) ] test :: (Int, String) test = $(fs (`id` (1,"2")) -- ryan On Sun, Jun 7, 2009 at 2:28 AM, Vladimir Reshetnikov<v.reshetnikov(a)gmail.com> wrote: > Hi Ryan, > > Thanks for your explanation. What language with dependent types would > you recommend me to look at? > > Now I am studying rank-N polymorphism in Haskell and trying to > generalize some combinators in my libraries to multitypes. This is how > I came to this question. > > Thanks, > Vladimir > > On 6/7/09, Ryan Ingram <ryani.spam(a)gmail.com> wrote: >> This is a really interesting question. >> >> So, fs is well-typed in Haskell: >> fs :: (((a,a) -> a) -> t) -> (t,t) >> i.e. >> fs id :: ((a,a) -> a, (a,a) -> a) >> >> However, I believe what you are asking is for fs to be equivalent to >> the following: >>> fs2 f g = (f fst, g snd) >> >> which has the type >> fs2 :: (((a, b) -> a) -> t) -> (((a1, b1) -> b1) -> t1) -> (t, t1) >> >> except with the argument broadcast polymorphically to both positions. >> >> This means the argument must have the multitype >> >> g :: ((a,b) -> a) -> t /\ ((a1,b1) -> b1) -> t1 >> >> for some t and t1 which are functions of a,b and a1,b1. >> >> Unfortunately I don't believe it is possible to encode this type in >> System F or System F(c), the underlying lambda-calculus used by GHC, >> so Haskell isn't going to be able to solve this problem. But there >> are statically typed languages which can solve this problem. >> >> You can take the big hammer of dependent types, and write fs something >> like this (not Haskell; this is a dependently-typed language): >> >> typeof_g :: (Type -> Type -> Type -> Type) -> Type >> typeof_g res_type = (a :: Type) -> (b :: Type) -> (c :: Type) -> >> ((a,b) -> c) -> res_type a b c >> >> fs :: (res_type :: Type -> Type -> Type -> Type) -> (g :: typeof_g >> res_type) >> -> (a :: Type) -> (b :: Type) -> (res_type a b a, res_type a b b) >> fs _ g a b = (g a b a fst, g a b b snd) >> >> So, you'd write fs id like this: >>> fs (\a b c. (a,b) -> c) (\a b c. id ((a,b) -> c)) >> >> This is a fascinating problem, though. What put you on this path? >> >> -- ryan >> >> On Sat, Jun 6, 2009 at 3:06 PM, Vladimir >> Reshetnikov<v.reshetnikov(a)gmail.com> wrote: >>> Hi, >>> >>> I have the following code: >>> >>> -------------------------------------------------------------------------------- >>> fs g = (g fst, g snd) >>> examples = (fs fmap, fs liftA, fs liftM, fs id, fs ($(1,"2")), fs >>> ((,)id), fs (:[]), fs repeat) >>> -------------------------------------------------------------------------------- >>> >>> The idea is that fs accepts a polymorphic function as its argument. >>> What type signature can I specify for f in order to compile this code? >>> If it is not possible in Haskell, is there another language with >>> static typing which allows this? >>> >>> Thanks, >>> Vladimir >>> _______________________________________________ >>> Haskell-Cafe mailing list >>> Haskell-Cafe(a)haskell.org >>> http://www.haskell.org/mailman/listinfo/haskell-cafe >>> >> >

1 0