- Haskell-Cafe - Haskell.org

Re: [Haskell] ANNOUNCE: GPipe-1.0.0: A functional graphics API for programmable GPUs
by Petr Pudlak 07 Oct '09

07 Oct '09

Hi Tobias, (I'm completely new to GPU programming, so my question may be completely stupid or unrelated. Please be patient :-).) Some time ago I needed to perform some large-scale computations (searching for first-order logic models) and a friend told me that GPUs can be used to perform many simple computations in parallel. Could GPipe be used for such a task? I.e. to program some non-graphical, parallelized algorithm, which could be run on a GPU cluster? Thanks for your answer, Petr On Sun, Oct 04, 2009 at 08:32:56PM +0200, Tobias Bexelius wrote: > I'm proud to announce the first release of GPipe-1.0.0: A functional graphics > API for programmable GPUs. > > GPipe models the entire graphics pipeline in a purely functional, immutable > and typesafe way. It is built on top of the programmable pipeline (i.e. > non-fixed function) of OpenGL 2.1 and uses features such as vertex buffer > objects (VBO's), texture objects and GLSL shader code synthetisation to create > fast graphics programs. Buffers, textures and shaders are cached internally to > ensure fast framerate, and GPipe is also capable of managing multiple windows > and contexts. By creating your own instances of GPipes classes, it's possible > to use additional datatypes on the GPU. > > You'll need full OpenGL 2.1 support, including GLSL 1.20 to use GPipe. Thanks > to OpenGLRaw, you may still build GPipe programs on machines lacking this > support. > > The package, including full documentation, can be found at: > http://hackage.haskell.org/package/GPipe-1.0.0 > > Of course, you may also install it with: > cabal install gpipe > > > Cheers! > Tobias Bexelius > > ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ > kolla in resten av Windows LiveT. Inte bara e-post - Windows LiveT är mycket > mer än din inkorg. Mer än bara meddelanden

2 1

better way to do this?
by Michael Mossey 07 Oct '09

07 Oct '09

I'm looking for a hint to write the following code with less redundancy. I have a constructor called BoxBounds, and I want to make one with random values. randomBox :: IO BoxBounds randomBox = do x <- getStdRandom (randomR (-10,10)) y <- getStdRandom (randomR (-70,70)) t <- getStdRandom (randomR (5,10)) b <- getStdRandom (randomR (5,10)) l <- getStdRandom (randomR (5,10)) r <- getStdRandom (randomR (5,10)) return (BoxBounds x y l r t b)

16 28

Network.Curl and posting XML data
by Erik de Castro Lopo 07 Oct '09

07 Oct '09

Hi all, I need to do a HTTP post of XML data (Content-type == "text/xml") to a HTTP (eventually HTTPS) server and to retrieve both the response code and some returned XML data. I have had an extensive read of the haskell-curl docs: http://hackage.haskell.org/packages/archive/curl/1.3.5/doc/html/Network-Cur… as well as playing around with some code, but haskell-curl doesn't seem to support this. I've tried curlPost which along with not returning a response also insists on encoding my XML data as application/x-www-form-urlencoded which simply doesn't work. Anyone ben able to get something like this to work? Are there any alternatives to haskell-curl which will work for my application? Cheers, Erik -- ---------------------------------------------------------------------- Erik de Castro Lopo http://www.mega-nerd.com/

4 8

Dutch HUG: meeting next week (October 12th) in Utrecht
by Tom Lokhorst 07 Oct '09

07 Oct '09

I'd like to invite functional programmers in The Netherlands to the Dutch Haskell User Group [1, 2] meeting next Monday at 19:00. Each meeting I see new faces, so that's why I continue to send these invitations to several mailing lists; it's working! At this meeting we'll have two talks. Andres Löh will talk about lhs2TeX, a tool for typesetting Haskell code using LaTeX. Yaakov Nemoy will talk about the status of Haskell packaging in Fedora and how to do it yourself. Like last month, the meeting will be in the Booth Hall [3] of the Utrecht University Library. There's free car parking [4] at the library, and it's easily accessible by bike and bus [5]. Unfortunately, I won't be able to attend, but I hope you will! - Tom Lokhorst [1]: http://www.haskell.org/haskellwiki/Dutch_HUG [2]: http://groups.google.com/group/dutch-hug [3]: http://www.uu.nl/EN/library/contact/university_library/zaalverhuur/Pages/de… [4]: http://www.uu.nl/EN/library/contact/university_library/Parkeren/Pages/defau… [5]: http://www.uu.nl/EN/library/contact/university_library/plattegrondenrou/Pag…

1 0

Creating an alias for a function
by michael rice 07 Oct '09

07 Oct '09

How do I create an alias for a function, like giving CAR the same functionality as HEAD. I know I can do it by creating a definition (see below), but is there a better way, like Scheme's (define head car) car :: [a] -> a car x = head x The reason for doing this is to more closely mirror legacy code. Michael

8 13

RETURNED MAIL: SEE TRANSCRIPT FOR DETAILS
by svens＠gmx.de 07 Oct '09

07 Oct '09

נ��55��7y�9(ߖ��! ��B��уN�(�a� h\H D�D Ĵ��׏��ɔ�2Qlo�q� ]u��\$`�.�,j~��E�M�{� ��;�{>v'�g��뱤�t��3D磻�y�b �hY((�[2TUm�1��[P�3�K��u��B�Uݠ�a�C�<�#�0�С�~�ŉ�|��8�l�~'Rf��("�LG�ǜ�2y��2��Zݑ�$Zss��Y51��V:�j�� [g1��9Z4o2��s�� laۛ�H!�ߤ4�&W�hM�4�S��"z��o}d��l:qi�D�H�$mRm3Vf�WG��'� �>��]�tA�� :Q�8��ɪ��yT[�!�)鶞i��3Go�R��㯂��R�.<�<ɗ��^e��.��.�b��Cj�� _��-3]��"��Q®��o_�"S_&�3U��"��i��3nҖ�h�B�j�3�ٺ�J �U�t�B��|<�m�y�XE� /a>"�>��e��Ȏe]��ʂ��嚕��Ei��Ocf��I$�דcM3��Ս5E�7��Y�$gV[�0~�Fo��4CA��/�h'z�BؿO��GM ��o, �]�[.�:�u'K�>Z��T�Eb̈́"ᧈk��9��_i\�!L�k�!9��Ki�H�탊e$&K�.*u��-�� f ��&_j\��p�"�Eq:<b0 M'�l�t�C��<�_��IV蓲QW��A��q� ��4��p��`��Z.'t[��c?��z�#�/[��u��hG��HG�� ą?iC�4 R2[X~�;,��12:P2 �U\i��*n�%�{ �"ZX^�W�Ⱦ� M��QZ��)�]�cՅi�X�`��S��h��LbI��#$cY~ЁB�k�E��*�dXK��~X�DļHRZɜ��SԨϘ�P�L�� :�4?�V�3<��9E�3�ϟY�(K��/�Y��ShD��ց��E��R�w}�a|e�&-��Q_P�^j��6�W}8*m��-��ڃ�ck3�/�,?�%h�h�2 ѝY�О��C��w��5�X� �MK U<�#�2<a T<zRe��91��>k�Ϛ� f�?u�� 6x8Y8F'�:> �0Zq�fh��A��U߲��7(�E�t�r(G��O �[8)mk6�_�$d��^k8�K��W젧�u/��҆�So�z6�[آ��r��f��u"�ae��-� �Oc��a72Z�jwz��gq�$�Q�~�-{��:g��!SX 2c�(��Zm~��Ǘ��U��K?�s�;�/G&��(��|�:��Rd�{��K��>��$N�e�EW9��0�qa��Jsy�!Ix�� wA#�[wq\o٘.��1�ZhIT� n8˖�I�� :��k<��`�Z�Ơ��#9��͜߆��L�jb��gM�e��R1N��ؘ͐�pN 6��o�;�X��O\/��d��*�L��ʁ ��Dݸ�&:$0��[a�x��vğC'Dua��ҕ�k �w� 2C�B��D�R�Va9]��#zn��n-K|�k�%��:�I��9J� 1:��ѡ>'�9�^��"Jo�~�dΐV�US��x}��B�lō�궼�^��\�ghM��pP\�#d�Z� ��9��*��k�#�P�d��*'��-�R�� iC�m�œﵜM�>`<n/E�� 8��;u��l�� CH�pմrHl��^w��B��_��e��,��iyv�c�3�vƵ�-L%�.��t~��W��Mo��7�-4�$�J��G"]�WS��8âp��9�}��A"�O�6N*��J��&.�:R�� /�uu�捎KϬ�(�EoS�To�tF'�zLz��:�o��'*2\O��2Iad�r��W�`hp�W]�SVʫ9��Lݼ��{&6��ժ#��(�݇��Jڿ�bM<��T� ţ�I�3ӃͲu�eW� \��?�!�� ş�$(��\%�(D�e�9��PX��x�AI�-j��J�SUi��o��qI�� Pp��PD�(Rΐ��g�d��mȇ��Vgu1:��Ż��04a�}"�-u3�"w̪��x��c�� g��g�p��5�,Ю��}�P�oK�H�x��T ;�k��̽�*�?��7�4��]/�94��$�֋ӘU�ޖHBe��p��<��9��˓!U��{�Aϻfb�y��stl��9, 5d�r"Q�&8f .d mcqݟv�.Hڈ��s� N��z�9i��R1��8~0.~��p��-#��v�ny�U��5ʪ��'��W�N��T�!�H3Rt oY�3��$��TV6�T/�,A�oDbQ�fO��X�赁w�r)��h �2��t��S�HQ�k��#m�0��M�R�x\��*��<� �e)�~��g��i��:��|toL��#J'tL��0�ڰ�.Ie�7��VZ�� `�rdg�K�H�v;�� <�~0d� D�7BC�R��P��\H4�7X��9�G(�e,�� "��T�M��Ԉ1��ꮞ��f_L� �勥��;�n$蛮��`��]��ǝtp�|[�{��b`�ɑ�K�bϋD#�Q�G;�x{��c��3�E��JV,�X�U��X?/j&Qt�G\��?o$.J �W�� !��'��enC��f�\��[U/��m& �� ԉ�]�o#}��㜥s{e��Q��zJ�, �0�YE��cRt��ӗ�CW�PD`��s4��&�A!��Uy�o�lB��>?��V��u %�;��s�^�䕘]��~�˫8.��{��|m��d-�AȪ�UqFF�� .��%��U䵛$��a�oK�*��n�n�|g�~��jI��2-u~�Ǳ��r��y�3և��98�i��da�cH�� ;h��3N�Jw$T�a��5ezD\z��x�z��7HM�� K8v&Mm�,v��ޚ�\��Ə��$p0<�mo�D��m#3}�f�\Q

1 0

Fwd: [Haskell-cafe] I read somewhere that for 90% of a wide class of computing problems, you only need 10% of the source code in Haskell, that you would in an imperative language.
by Alberto G. Corona 07 Oct '09

07 Oct '09

May be because consciousness is relatively new and thus, not optimized. Sequentiallity is somehow related with lack of information and lack or resources. There is nothing more sequential than a Turing machine. The Von Newman architecture is designed to make as much as possible with a few more resources. The older parts of the brain are fast and parallel because they have a long history of optimizations by natural selection., The new cortex still struggles, step by step, to deduce new information from the available internal and external information at the pace they arrive. 2009/10/1 Tom Tobin <korpios(a)korpios.com> On Thu, Oct 1, 2009 at 3:26 AM, Andrew Coppin > <andrewcoppin(a)btinternet.com> wrote: > > It might be a better argument to say that human thinking is fundamentally > > sequential; parallel computers have been around for a little while now... > > Perhaps *conscious* human thinking is sequential — yet our brains are > massively parallel processors, and have been around for quite a long > time. ;-) > _______________________________________________ > Haskell-Cafe mailing list > Haskell-Cafe(a)haskell.org > http://www.haskell.org/mailman/listinfo/haskell-cafe >

2 1

Fast times at Inferable-but-not-Checkable High.
by Joe Fredette 07 Oct '09

07 Oct '09

So, I've been fiddling with an utterly random idea. What if I had a class: class Hom a b where data Rep a b hm :: Rep a b -> b im :: a -> Rep a b That is, all types that have some conversion between them (an isomorphism originally, then I thought homomorphism, now I'm not sure what the hell I'm talking about, so I've just stuck with Hom...) which preserves some sense of "structure" and "content", for instance data CouldBe a = Only a | Nada is _obviously_ the same type as `Maybe`. However, it can't be used transparently where `Maybe` can be used. That is, shouldn't I be able to define a 1-1, onto function `phi :: CouldBe a -> Maybe a` and as such another `pho :: Maybe a -> CouldBe a` such that phi . pho = pho . phi = phum ... errr. `id`? Hom a b represents one end of that (specifically `hm . im = phi`, and `hm . im` for the instance `Hom b a` would `pho`), then I could, instead of writing a type which expects maybe, simply expects anything _equivalent_ to a maybe, eg safeHead :: Hom (Maybe a) b => [c] -> Rep a (Maybe c) safeHead [] = im Nothing safeHead (x:_) = im (Just x) Though- I think this is a little bit off in terms of how it should work, perhaps this is my problem, but the idea is that instead of returning a Maybe, it returns something which can be converted _from_ a maybe. That is, a "generic" type like: data X1 a = X a | Y which is the "form" of any "maybe-like" type. In one sense, I'm almost trying to be polymorphic over the constructors of a given type, kindof, sortof, if you squint your eyes just right and try not to think too much. My problem comes when I try to do this: hom = hm . im eq x y = hom x == hom y Which, I reason, ought to satisfy the type: eq :: (Hom a b, Eq b) => a -> a -> Bool this assumes that hom defines a equality-preserving conversion. However, the type it infers is: eq :: (Hom a b, Hom a1 b, Eq b) => a -> a1 -> Bool Now, this makes sense, all my signature is is a special case of this type. Interesting consequence, this eq could compare a `Maybe a` and `CouldBe a` in a sensical way. Which is a nice benefit. However, here's where it gets weird, If I try to _provide_ this signature (the second, more general one, or the first, specific one, it doesn't matter), GHC gives me the following error: Iso.hs:29:10: Could not deduce (Hom a b) from the context (Hom a b1, Eq b1) arising from a use of `hom' at Iso.hs:29:10-14 Possible fix: add (Hom a b) to the context of the type signature for `eq' In the first argument of `(==)', namely `(hom x)' In the expression: (hom x) == (hom y) In the definition of `eq': eq x y = (hom x) == (hom y) Failed, modules loaded: none. for the latter and a similar one (but for each of the `Hom a b`, `Hom a1 b` cases. Punchline is this, I'm trying to write a generic equality that works for any two types with the same constructor-structure, when I try to write said function, GHC can infer but not Check (it seems, I'm not sure if that's the correct lingo) the type... My question is twofold. 1. Is this whole idea even possible? I mean- is there some limitation I'm going to run into, I have run into problems when trying to do: instance (Hom a b, Eq b) => Eq a where blah wrt UndecideableInstances. But it seems so _obvious_ what that means, I don't think I fully understand what's going on -- that is, whether I'm misunderstanding how class contexts work here, and why this (seemingly obvious) signature is undecidable, or whether I'm just mistaken in my type signature all together. (My goal is to say that anything which can be converted freely to and from something that is an instance of the Eq class, must preserve the equality relationships. 2. Why is GHC unable to check the types, but infer them. I always understood inferring to be the "hard" part of the problem, and checking to be the "easy" part. Is my intuition wrong? I've posted the code here[1] for your perusal. It's not particularly important that I get this solved, it's just a random idea I wanted to explore and thought the notion might appeal to some other Haskeller's around there who like to have self-documenting datatypes, but hate having to rewrite lots of simple utility functions as penance for our documentarian sins. /Joe

2 1

Fast times as Inferable-but-not-Checkable High -- Link to Source
by Joe Fredette 07 Oct '09

07 Oct '09

Sorry, I forgot to actually include the link to the source: http://lowlymath.net/Iso.hs /Joe

1 0

Fwd: [Haskell-cafe] Num instances for 2-dimensional types
by Alberto G. Corona 07 Oct '09

07 Oct '09

Why? real numbers, complex nuimbers, n-dimensional spaces have well defined + and * operations (vectorial product in the latter case). even algebraic expressions like: data Expr = Var String | Number Integer | Sin Expr | Cos Expr can be instances of Num and express certain simplification rules in the definition. instance Num Expr where ... (Sin x) * (Cos x) = (Sin (2 * x))/2 ... I started to develop a extensible symbolic math manipulation around the idea of using symbolic expressions as instances of Num and other basic classes. But I leaved it out due to lack of time. By the way I attach it just in case anyone want to get some idea from it: module DynAlgebra where data Dyn= forall a.(Num a, Eq a, Show a, Ord a, Read a, Simplify a)=> Dyn a instance Show Dyn where show (Dyn a)= show a instance Eq Dyn where (Dyn a) == (Dyn b)= solve a== solve b class Simplify a where simplify :: a -> a solve :: a -> Polonomial Expr1 data Polonomial e = e :/ e | e :+ e | e :* e deriving Show data Expr1= Var String | I Integer | Pi | E deriving Show -- Integer :/ Integer for exact rational arithmetic data Expr= Polinomial Expr | Formula Dyn deriving (Show, Eq) instance Num (Polonomial Expr1) where fromInteger = I (+) (I a) (I b)= I (a+b) (+) e1 e2 = e1 :+ e2 (*) (I a) (I b)= I (a*b) (*) e1 e2 = (:*) e1 e2 abs (I x) | x>=0 = I x | otherwise = I (-x) signum (I x) | x >= 0 = 1 | otherwise= -1 instance Ord Polonomial where compare (I a) (I b)= compare a b compare ( x :+ z) y | x < y && z >0 = LT | x > y && z >0 = GT compare y ( x :+ z) | y > x && z >0 = LT | y > x && z >0 = GT instance Eq Polonomial where (I a) == (I b) = a == b (I a :+ I b)== ( I c :+ I d)= a + b== c + d (I a :* I b)== ( I c :* I d)= a * b== c * d (a :+ b)== (c :+ d)= a==c && b == d (a :* b)== (c :* d)= a== c && b == d (a :/ b)== (c :/ d)= a * d == b * c exp1 == exp2 = simplify exp1== simplify exp2 instance Simplify (Polonomial Expr1) where solve x = simplify x simplify ( (I x) :/ (I y))= case quotRem x y of (q,0) -> I q (q,r) -> let m= mcd y r in (I(x `div` m)) :/ (I(y `div` m)) simplify ((I a) :+ (I b))= I (a + b) simplify ((I a) :* (I b))= I (a * b) simplify ((a :* b) :+ (c :* d)) | a == c = simplify $ a * (b :+d) simplify (exp1 :+ exp2) = simplify exp1 :+ simplify exp2 simplify (exp1 :* exp2) = simplify exp1 :* simplify exp2 simplify expr= expr mcd x y= case mod x y of 0 -> y t -> mcd y t subst:: Polonomial -> [(String, Polonomial )] -> Polonomial subst exp l= subs1 exp where subs1 (Var v)= case lookup v l of Nothing -> Var v Just e -> e subs1 (e1 :+ e2) = ((subs1 e1) :+ (subs1 e2)) subs1 (e1 :* e2) = ((subs1 e1) :* (subs1 e2)) subs1 (e1 :/ e2) = ((subs1 e1) :/ (subs1 e2)) subst e= e f x= x :* x main= print $ solve $ 2 :+1 2009/10/5 Miguel Mitrofanov <miguelimo38(a)yandex.ru> > > Sönke Hahn wrote: > > I used to implement >> >> fromInteger n = (r, r) where r = fromInteger n >> >> , but thinking about it, >> fromInteger n = (fromInteger n, 0) >> >> seems very reasonable, too. >> > > Stop pretending something is a number when it's not. > > _______________________________________________ > Haskell-Cafe mailing list > Haskell-Cafe(a)haskell.org > http://www.haskell.org/mailman/listinfo/haskell-cafe >

1 0