May 2008 - Haskell-Cafe

ANN: xmonad-utils 0.1
by Gwern Branwen 11 May '08

11 May '08

Just dropping in a quick note: I've uploaded to Hackage the 0.1 release of xmonad-utils here: <http://hackage.haskell.org/cgi-bin/hackage-scripts/package/xmonad-utils>. There is of course a Darcs repo available as well: <http://gorgias.mine.nu/repos/xmonad-utils>. What is xmonad-utils? It's a couple of small Xlib programs which might be useful for Xmonaders. In particular: * hxsel can replace xclip, and was the foundation for the XMonad.Util.XSelection module in XMonadContrib/; it gets a copy of the copy-paste buffer to manipulate * hhp is a replacement for unclutter, the famous utility which hides your mouse when not in use and so cleans up your screen (but which is also famously unmaintained and buggy). I've personally found hhp to be very helpful, so I packaged it all up. Miscellaneous notes: * There is also a 'heval' program, which tries to provide an X prompt to a GHC eval through the GHC API, but that may or may not work for you (it doesn't compile for me, but I'm fairly sure that's because my 'directory' package is b0rked in an unrelated manner). * There is a requirement on cabal >= 1.4; this was needed because older Cabals didn't properly include in the sdist tarball one of the .hsc files. I don't believe it's actually needed to build. -- gwern nuclear Monarch NSADS Scud foundation World or Kilo SADMS M-x

1 0

Cryptographic hash uniquness (was [Haskell-cafe] Simple network client)
by Peter Verswyvelen 11 May '08

11 May '08

>winds up having a write cache, which is mutable in practice. The >interesting thing is that the block's location is the cryptographic >hash of its contents, which leads to all sorts of neat properties (as >well as requiring immutability). That's interesting. When I developed a version control system for a customer, I also used a cryptographic hash as the database key of file+content in question, but I was afraid I might have clashes (two files with different content generating the same hash)... My intuition told me that the odds of two cryptographic hashes (on meaningful content) colliding was much less than the earth being destroyed by an asteroid... But this is just intuition... What does computer science tell us about this? Thank you, Peter

6 6

Error: Improperly terminated character constant
by Ivan Amarquaye 11 May '08

11 May '08

I'm writing a function dRop to accept words ending in 'es' and drop the last two characters i.e. 'es'.eg. mangoes -> mongo but i keep on getting this error: "Improperly terminated character constant" after running this code which i have left below. Can i get any form of help from anyone in here...? ----------------------------------------- dRop :: String -> String dRop word = if drop (length word - 2) word == 'es' then take (length word - 2) word else word -------------------------------------------------- regards Amarquaye.Ivan _________________________________________________________________ Connect to the next generation of MSN Messenger http://imagine-msn.com/messenger/launch80/default.aspx?locale=en-us&source=…

2 1

Newbie Question: Using Haskell Functions in a C Program
by Philip Müller 11 May '08

11 May '08

Hi, I'm in the process of writing a C program, but I can't stop thinking about how some functions would be much nicer implemented in Haskell. Is there a way to write some of the functions in Haskell and then use them in my C code via some kind of interface? BTW yes, I have been thinking about writing the whole program in Haskell, but I just don't have the level of experience in Haskell programming for that, since it's really heavy on IO. Thanks in advance for the answers! Regards Philip

5 7

Order of Evaluation
by PR Stanley 11 May '08

11 May '08

Hi (take 4 . map (>0)) (f s t) where s = 2 : t t = 3 : s f = zipWith (-) What would be the order of evaluation for the above code? How would I illustrate the evaluation step-by-step? I'm guessing that the code necessitates lazy evaluation and as such it starts with take then it applies f which in turn applies s and t and zipWith until the first element satisfies the predicate in map and This is repeated 4 times What does the list think? Many thanks, Paul P.S. I'm not done with induction. I'm just letting it rst for a bit.

10 14

Stack vs Heap allocation
by Edsko de Vries 11 May '08

11 May '08

Hi, How can I know whether something will be stack or heap allocated? For example, in the standard example of why foldl (+) 0 will fail to evaluate a long list of integers due to a stack overflow, but foldl' won't, it is pointed out that foldl starts building up unevaluated thunks. So, apparently, these thunks are allocated on the stack rather than on the heap with a pointer to the thunk on the stack. (I understand that foldl' is asymptotically better than foldl space-wise; that is irrelevant to my question.) On the other hand, in this version that sums all the values in a tree data Tree = Leaf Int | Node Tree Tree sum :: Tree -> Int sum t = sum' [t] 0 where sum' [] acc = acc sum' (Leaf i : ts) acc = sum' ts $! (i + acc) sum' (Node l r : ts) acc = sum' (l : r : ts) acc we are building up a (potentially) large lists of trees yet to be processed, but never run out of stack space. Apparently, the list is built up on the heap rather than on the stack. What is the fundamental difference between these two examples? Why is the list of trees yet to be processed allocated on the heap (with a pointer on the stack, presumably) but the unevaluated thunks in the foldl example allocated on the stack? Thanks, Edsko

5 6

seeking advice on my POSIX wrapper
by Galchin, Vasili 10 May '08

10 May '08

Hello, Last night I sent out an announcement about some POSIX work that I have been doing. In any case, one of the FFI wrappers is driving me crazy, i.e. the one for mq_receive: http://opengroup.org/onlinepubs/007908799/xsh/mq_receive.html . When I call this function (mqReceive), I get "message too long". In my test cases I am sending and receiving messages that are only 11 bytes! The wrapper seems really straightforward. Perhaps I am looking right at the problem and don't see. I need other eyes on the wrapper to help me ;^). Please see below. Regards, V. ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- -- mqReceive is still being debugged!!!!!!!!!! -- | Retrieve a message from mqueue designated by "mqd" -- mqReceive :: Fd -> ByteCount -> Maybe Int -> IO (String, Int) mqReceive (Fd mqd) len (Just prio) = do allocaBytes (fromIntegral len) $ \ p_buffer -> do with (fromIntegral prio) $ \ p_prio -> do rc <- throwErrnoIfMinus1 "mqReceive" (c_mq_receive mqd p_buffer (fromIntegral len) p_prio) case fromIntegral rc of 0 -> ioError (IOError Nothing EOF "mqReceive" "EOF" Nothing) n -> do s <- peekCStringLen (p_buffer, fromIntegral n) return (s, n) mqReceive (Fd mqd) len Nothing = do allocaBytes (fromIntegral len) $ \ p_buffer -> do rc <- throwErrnoIfMinus1 "mqReceive" (c_mq_receive mqd p_buffer (fromIntegral len) nullPtr) case fromIntegral rc of 0 -> ioError (IOError Nothing EOF "mqReceive" "EOF" Nothing) n -> do s <- peekCStringLen (p_buffer, fromIntegral n) return (s, n) foreign import ccall unsafe "mqueue.h mq_receive" c_mq_receive :: CInt -> Ptr CChar -> CSize -> Ptr CInt -> IO CInt

3 4

Old editions of The Monad.Reader lost
by Henk-Jan van Tuyl 10 May '08

10 May '08

L.S., Now that all hawiki pages have been removed, we have lost some valuable information. For example The Monad.Reader; on http://www.haskell.org/haskellwiki/The_Monad.Reader it says: Older editions can be found on the old Haskell wiki – they haven't been included here for licensing reasons. The page it links to, a hawiki page, has dissappeared. I propose to bring at least the The Monad.Reader pages back. Backups of these pages can be found at the Wayback Machine, but there is no guarantee that will always be there. -- Met vriendelijke groet, Henk-Jan van Tuyl -- http://functor.bamikanarie.com http://Van.Tuyl.eu/ --

3 6

Type unions
by Eric Stansifer 10 May '08

10 May '08

I have been trying to write a DSL for Povray (see www.povray.org) in Haskell, using the technique of: http://okmij.org/ftp/papers/tagless-final-APLAS.pdf with some inspiration taken from http://okmij.org/ftp/Haskell/DSLSharing.hs The Povray Scene Description Language is a very declarative language, with few high level constructs (even loops take a bit of work) -- which is why I'm putting it in Haskell. At one point, I needed a "varargs" function for the DSL, a function f :: b -> a -> b dressed up to take a variable number of 'a's, known at compile time. This was easy enough: > data Nil a = Nil > data Cons b a = a ::: b a > infixr 1 ::: > > class VarArgs v where > apply_args :: (s -> a -> s) -> s -> v a -> s > > instance VarArgs Nil where > apply_args _ start _ = start > > instance VarArgs b => VarArgs (Cons b) where > apply_args f start (a ::: b) = apply_args f (f start a) b The solution is quite workable: I can simply write the following, and I believe the summation is expanded out at compile-time: > apply_args (+) 0 (2 ::: 3 ::: 8 ::: 1 ::: (-3) ::: Nil) But I found I also needed a function to take a union type -- that is, the function would either take an argument of type T1, or of type T2, known at compile time. I tried a similar technique as I tried with varargs, and unfortunately ended up with this: > data LeftOf a b = L a > data RightOf a b = R b > > class Union u where > apply_union :: (a -> c) -> (b -> c) -> (u a b) -> c > > instance Union LeftOf where > apply_union f _ (L a) = f a > > instance Union RightOf where > apply_union _ g (R b) = g b > > type A = Integer > type B = String > type C = () > > type Union_ABC u1 u2 = u1 A (u2 B C) > > f_A = show . (+ 3) > f_B = reverse > f_C = const "unit" > > f :: (Union u1, Union u2) => Union_ABC u1 u2 -> String > f = apply_union f_A (apply_union f_B f_C) > > main = do > putStrLn $ f $ (L 6 :: Union_ABC LeftOne LeftOne) > putStrLn $ f $ R (L "hello, world") > putStrLn $ f $ R (R ()) Notice a lot of ugliness in my example: e.g., the definition of f, the type signature of f (I can't move the context into the type-synonym Union_ABC), creating objects of the union type, and the unpleasant surprise that I needed to provide the type of 'L 6'. This solution is very not-scalable: the Povray SDL is a "messy" language, and for my DSL I would need approximately 20 or 30 such unions, each a union of about 20 types (admittedly with a lot of overlap from union to union). I think the solution is to abandon the lofty ideal of statically determining argument types; instead have a universal type with tags to distinguish types dynamically: > data Universal = UA A | UB B | UC C > f :: Universal -> String > f (UA a) = f_A a > f (UB b) = f_B b > f (UC c) = f_C c > > main2 = do > putStrLn $ f $ UA 6 > putStrLn $ f $ UB "hello, world" > putStrLn $ f $ UC () ...but I'm not ready to give up hope yet. Suggestions please? Eric

3 3

I am new to haskell
by Ambrish Bhargava 10 May '08

10 May '08

Hi All, I am new to Haskell. Can anyone guide me how can I start on it (Like getting binaries, some tutorials)? Thanks in advance. -- Regards, Ambrish Bhargava

8 11