March 2007 - Haskell-Cafe

"runST $ return ()" /= "runST (return ())" ??
by Marc A. Ziegert 24 Mar '07

24 Mar '07

hi! i've just discovered this strange behaviour of existential quantifiers with runST: --- Prelude Control.Monad.ST> :t runST (return ()) runST (return ()) :: () Prelude Control.Monad.ST> :t runST $ (return ()) <interactive>:1:9: Couldn't match expected type `forall s. ST s a' against inferred type `m ()' In the second argument of `($)', namely `(return ())' Prelude Control.Monad.ST> --- the same with "id runST undefined". is this a bug or an unsolved problem? i'm not sure wheather it is a part of the wanted feature of runST's type definition. i did not find any discussions about this. can anyone enlighten me, please? - marc

2 2

OpenGL and GLUT in GHC
by Ruben Zilibowitz 24 Mar '07

24 Mar '07

Hi, I'm experimenting using OpenGL and GLUT in Haskell using GHC. There are modules Graphics.Rendering.OpenGL and Graphics.UI.GLUT. I am using these. I've encountered a strange bug which I'm having trouble with. If I render a sphere and a plane, then the plane is facing the wrong way and is shaded on the wrong side. If however I only render the plane then it appears to be facing the right way and is shaded on the correct side. I have made the source file available as a download here: http://www.cse.unsw.edu.au/~rubenz/stuff/test.hs It can be built by running: ghc --make test.hs I am using ghc 6.6 There is a comment in the source file saying: "Commenting out the line below here causes the plane to be rendered facing towards the camera..." The bug can be seen by commenting out the line of code that follows and recompiling. If anyone can help me by explaining why I am getting this bug or how to fix it that would be great. I'd be very appreciative. Cheers, Ruben

2 1

Performance Help
by Dominic Steinitz 24 Mar '07

24 Mar '07

I have re-written the sha1 code so that it is (hopefully) easy to see that it faithfully implements the algorithm (see http://www.itl.nist.gov/fipspubs/fip180-1.htm) Having got rid of the space leak, I have been trying to improve performance. Currently, the haskell code is 2 orders of magnitude slower than the sha1sum that ships with my linux. > dom@heisenberg:~/sha1/testdist/sha1> time ./perfTest perfTest > c7eae62ddabb653bb9ce4eb18fa8b94264f92a76 > Success > > real 0m2.152s > user 0m2.112s > sys 0m0.028s > dom@heisenberg:~/sha1/testdist/sha1> time sha1sum perfTest > c7eae62ddabb653bb9ce4eb18fa8b94264f92a76 perfTest > > real 0m0.057s > user 0m0.008s > sys 0m0.004s I've played around with profiling and doubled the performance of the haskell code but I'm nowhere near the C performance. > Sun Mar 11 19:32 2007 Time and Allocation Profiling Report (Final) > > perfTest +RTS -p -RTS eg > > total time = 6.75 secs (135 ticks @ 50 ms) > total alloc = 1,483,413,752 bytes (excludes profiling overheads) > > COST CENTRE MODULE %time %alloc > > oneBlock Data.Digest.SHA1 39.3 40.1 > $& Data.Digest.SHA1 20.7 21.6 > f Data.Digest.SHA1 13.3 6.2 > getWord32s Data.Digest.SHA1 7.4 6.6 > test2 Main 5.9 8.7 > blockWord8sIn32 Data.Digest.SHA1 5.2 5.3 > blockWord8sIn512 Data.Digest.SHA1 3.0 4.4 > pad Data.Digest.SHA1 1.5 3.5 > k Data.Digest.SHA1 1.5 0.0 > fromBytes Data.Digest.SHA1 0.0 3.5 Here's the code that is taking the majority of the time. > ($&) :: [Word32] -> [Word32] -> [Word32] > a $& b = zipWith (+) a b > > -- Word128 -> Word512 -> Word128 > oneBlock ss xs = Word128 (as!!80) (bs!!80) (cs!!80) (ds!!80) (es!!80) > where > ws = xs ++ map (rotL 1) (zipWith4 xxxor wm3s wm8s wm14s ws) > where > xxxor a b c d = a `xor` b `xor` c `xor` d > wm3s = drop (16-3) ws > wm8s = drop (16-8) ws > wm14s = drop (16-14) ws > as = ai:ts > bs = bi:as > cs = ci:(map (rotL 30) bs) > ds = di:cs > es = ei:ds > ts = (map (rotL 5) as) $& (zipWith4 f [0..] bs cs ds) $& es $& (map k > [0..]) $& ws > Word128 ai bi ci di ei = ss Any help would be appreciated. I've put a copy of a working system here if anyone wants to experiment (http://www.haskell.org/crypto/downloads/sha1.tar.gz) Thanks, Dominic.

7 10

Rank-2-polymorphism problem
by Martin Huschenbett 24 Mar '07

24 Mar '07

Hi, I'm writing some database code using HSQL and had to stop on a problem with rank-2-polymorphism that I can't solve. The essence of my code is: module Value where import Data.Maybe class SqlBind a where fromSqlValue :: String -> a data Field data Value emptyValue :: Field -> Value emptyValue _ = ... readValue :: Field -> (forall s. SqlBind s => s) -> Value readValue _ = ... That works just fine. But now I want a version of readValue that has a Maybe wrapped around the second parameter and that shall call readValue in the case of a Just and emptyValue in the case of Nothing. But I can't figure out how to write this function as I always get compiler errors. My trials were: -- The type I want to get. readValue' :: Field -> (forall s. SqlBind s => Maybe s) -> Value -- First trial: readValue' fld s = if isJust s then readValue fld (fromJust s) else emptyValue fld -- Second trial: readValue' fld s | isJust s = readValue fld (fromJust s) | otherwise = emptyValue fld -- Third trial: readValue' fld (Just s) = readValue fld s readValue' fld Nothing = emptyValue fld -- Fourth trial: readValue fld s = case s of Just s' -> readValue fld s' Nothing -> emptyValue fld But none of these trials worked. Is there any solution that works with GHC-6.6 for now? Thanks in advance, Martin.

5 8

Re: Bouncing ball in Gtk2Hs
by Duncan Coutts 23 Mar '07

23 Mar '07

On Fri, 2007-03-23 at 20:24 +0300, Thach Si Lam wrote: > I'm trying to write Bouncing Ball demo in Gtk2hs. But i coudn't find > timer class (like in wxhaskell demo) in Gtk2hs. There is timeoutAdd: http://haskell.org/gtk2hs/docs/gtk2hs-docs-0.9.10/System-Glib-MainLoop.html… used like so: timeoutAdd action 1000 where action is an action with type IO Bool and the number is the time to wait in milliseconds, (ie 1000 is 1 second). If the action returns True then it will get called again at the next timeout, otherwise it will not be called again. > How to deal with it ? In fact i have 2 processes: p1 for computing > some variables and p2 for changing (applying those variables) in > canvas. How to let them communicate together and update to canvas ? > Please, help me. Thank you very much. I'm forwarding this to the Haskell Cafe mailing list where someone might be able to give you a pointer. Duncan

1 0

Re: [Haskell-cafe] AI Strike Force!
by Andrzej Jaworski 23 Mar '07

23 Mar '07

Hi Andrew and all the involved, The idea is great. It will allow programming with a rewording feeling of depth to the exercise right from the start. However tackling some of the topics you mentioned, like GA, one first needs to develop new solid programming techniques that would circumvent Haskell's inbuilt reluctance to update variables distractively. I would myself wish to read a well documented study of using ST monad or finite maps to achieve this end. Perhaps then, your page should openly encourage the need to develop such "brute force" methods. Cheers, -Andrzej

1 0

HDBC Windows binaries with SQLite 3 support
by Alex Queiroz 23 Mar '07

23 Mar '07

Hallo list, I guess the subject line says it all. Can anybody help me? -- -alex http://www.ventonegro.org/

4 7

Last chance to join the Summer of Code!
by Malcolm Wallace 23 Mar '07

23 Mar '07

Haskell.org is currently looking rather short of student proposals for Summer of Code 2007 compared to last year. If you haven't yet thought about submitting an application, why not consider it now? You could be paid $4500 for three months work June-Aug! Apparently, the more proposals we get, the more places are likely to be funded. Also, Google have announced a relaxation of the deadline for student proposals. The new deadline is Mon 26th March at 1700 Pacific Standard Time (Tue 27th, 0000 UTC), so you have all weekend to get writing! Regards, Malcolm

1 0

ghc warning "Var/Type length mismatch"
by Nicolas Frisby 23 Mar '07

23 Mar '07

When I load my program, GHC spits these messages at me, but doesn't fail Any idea what might be causing this or how to figure that out? Var/Type length mismatch: [] [a{tv aGIf} [tau]] ... Var/Type length mismatch: [] [a{tv aGN8} [tv]] ... I found the responsible code in GHC's darcs, but the context didn't lend any help my feeble non-GHC hacker brain. My program is kind of big and I'm using personal libraries that exercise the type system a lot, but I'll wait to see if anyone is interested or if this has been handled before I share the gory details. This tidbit might help: I have a couple of usages of undefined in my program b/c I'm just starting to code and if I replace a particular undefined of this with some typed dummy code, it removes the second set of mismatches, but not the first. I tried also replacing the other undefined, but the first set of mismatches remained. Glasgow Haskell Compiler, Version 6.7.20070214, for Haskell 98, compiled by GHC version 6.7.20070214 Also, is there a better list for this question? I half-heartedly tried haskell.org's list of mailing lists for an appropriate place for GHC bugs but didn't find one. Thanks for your time, Nick

4 3

type refinement to a typeclass constrained type
by Daniil Elovkov 23 Mar '07

23 Mar '07

Hello folks A section on GADTs in the GHC user's guide gives a simple example: data Term a where Lit :: Int -> Term Int Succ :: Term Int -> Term Int IsZero :: Term Int -> Term Bool eval :: Term a -> a eval (Lit i) = i eval (Succ t) = 1 + eval t eval (IsZero t) = eval t == 0 let's add Sum :: Term Int -> Term Int -> Term Int eval (Sum l r) = eval l + eval r it's ok But when I do this Sum :: (Num a) => Term a -> Term a -> Term a eval (Sum l r) = eval l + eval r --- same it doesn't typecheck with the error "no isntance Num a" writing eval (Sum (l::Num b => Term b) (r::Num b => Term b)) doesn't help. I wonder, if there's a way to do what I'm trying to do. After all, I'm allowed to declare Term that way. Or is this a restriction of type refinement?

2 5