- Haskell-Cafe - Haskell.org

"Rewrite thunk action" rule?
by Peter Todd 22 Dec '08

22 Dec '08

I have a program with this data structure: data Element = Element { elementOrigin :: V, elementSubs :: [Element] } and this important bit of code that operates on it: transform :: T -> Element -> Element transform t e = Element { elementOrigin = tmulv t (elementOrigin e), elementSubs = map (transform t) (elementSubs e) } Where V is a vertex and T is a matrix transformation. The following is true: transform a (transform b e) == transform (a * b) e I have read about rewrite rules, such would efficiently rewrite the obvious transformations. However, given that the program will be applying *many* transformations to very deep and wide Element trees, and applying those transformations at many levels, I don't see how the optimizer would be able to handle all cases, for instance a thunk getting left over from perhaps evaluation prior to some IO function. FWIW here's an example "tree-builder" I'm using to test with: mkElems 0 _ = Element { elementOrigin = V 0 0, elementSubs = [] } mkElems depth width = Element { elementOrigin = V 0 0, elementSubs = replicate width $ transform (rotation (pi / (fromIntegral width))) $ transform (translation $ V 1 1) $ mkElems (depth - 1) width } with depth = width = 5 If I could somehow arrange for the transform function to detect when it is being applied to a unevaluated thunk, and then modify the thunk in place, that would basically be the behavior I need. Any suggestions? -- http://petertodd.org 'peter'[:-1]@petertodd.org

3 4

Re: [Haskell-cafe] A hell of a question
by Andrzej Jaworski 22 Dec '08

22 Dec '08

Arnaud Bailly kindly exposed my mistake: I cited Leibniz Monadology attributing it to Spinoza. Call me idiot but it wasn't thoughtless mistake, I really mixed up Spinoza's concept of 'modes' with Leibniz's concept of 'monad'. But it is because of my laziness rather than foolishness (at least this time). My argument is that in his final writings Leibniz's 'monads' encompass reality as perceptions and emanate from God as thought emanates from the mind, which is exactly what Spinoza's 'modes' are about. So monads and modes being equal I prefer Spinoza as a patron for Haskell for his purity of cause-effect treatment. Arnold adds another Haskell-Spinoza nicety, I hope he won't mind my including his all letter. Thank you Arnold, it is encouraging to find guys like you. Philosophy is not my thing but I believe every man should tackle it as exercise. Hello Andrzej, Thanks a lot for your bit of rationalism among all these devotions :-) I think however that you are confusing Spinoza with Leibniz in the following assertion. Monads are a concept invented by the latter, together with the famous "the best possible world ever" mocked by Voltaire in "Candide". The comparison between Haskell and spinozism is rather interesting though, especially when one considers that Spinoza's "Ethic" is based on the idea that the ultimate goal of one self is to increase its power to live, an affect which it calls "Joy". Thinking of Haskell as a way to increase one's joy and one's power is a nice thought. Regards, Arnaud Bailly

1 0

Re: [Haskell] ANN: HLint 1.0
by Gwern Branwen 22 Dec '08

22 Dec '08

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA512 So has anyone figured out how to integrate hlint with ghci? I've tried several approaches. Using hlint as a preprocessor ( http://www.haskell.org/ghc/docs/latest/html/users_guide/options-phases.html… ) doesn't work because the pre-processor needs to take 3 args. Even if one writes a wrapper shell script along the lines of 'cp $2 $3 && hlint $2', this still won't work. It apparently breaks when you load a file which requires another file. I am not sure why. I then thought to redefine :load and :reload in ghci; this would work very well with Emacs and Yi, since they tell ghci to :load and :reload buffers. Turns out, you're not allowed to overwrite :load and :reload. You can do :def reload whatever to your heart's content, but the new version never takes. (Not even if defined in .ghci.) Alright. A :hlint would probably be enough. I can always edit Yi/Emacs to follow a :reload/:load with a :hlint, after all. And aren't there shell command primitives already? So I set out, and I realize: I can't simply settle for some version of :! "hlint .", because the working directory rarely changes - if one opens ghci in ~/, and loads a file elsewhere, :pwd reveals one to still be in ~/. And a 'hlint .' could take a very long time indeed in ~/. Well, alright. I just need the file name of the module I'm working with. Surely there is an easy way to do that. And indeed, :show modules gives me what I'm looking for: *Recorder Control.Monad Data.Char Data.List> :show modules Util ( Util.hs, interpreted ) Recorder ( Recorder.hs, interpreted ) Game ( Game.hs, interpreted ) Monadius ( Monadius.hs, interpreted ) Demo ( Demo.hs, interpreted ) Well, sort of. Ok, we can parse that. Let's assume a variable x holds the output of :show modules as a String. We call lines on it, then map words on it, do a !! 2 on it, and we get ["Util.hs,", "Recorder.hs,", "Game.hs,", "Monadius.hs,", "Demo.hs,"]. Chuck in a map (filter (\= ',')), and we get a good list. We can turn the list into a string suitable for hlint with a quick unwords. So our long sought after command becomes ':def hoogle (\_ -> return $ ":! " ++ (unwords $ map (filter (\= ',')) $ (map words $ lines x) !! 2))'. But wait, how do we get 'x'? How do we call :show modules inside a Haskell expression? I have carefully looked over http://haskell.org/haskellwiki/GHC/GHCi#Using_GHCi and http://www.haskell.org/ghc/docs/latest/html/users_guide/ghci-commands.html and my conclusion is that you can't. You can't do a let x = :show modules, there is no function which will take ":show modules", and so on. :functions can accept Haskell output, but it's a one-way barrier. It's no good writing Haskell functions which need information from the :functions. Which all leaves me where I started - no ghci integration with hlint. Any ideas? - -- gwern -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.9 (GNU/Linux) iEYEAREKAAYFAklNNSIACgkQvpDo5Pfl1oJwWgCfZJ7nwkHfsZymIRass0ewmhFE en4AnjmQam3V8Go6pGoLTRsp5zZEzAms =nmJm -----END PGP SIGNATURE-----

2 1

A hell of a question
by Andrzej Jaworski 22 Dec '08

22 Dec '08

Hi Luke, When neurosurgeons split the brain into left and right hemisphere cutting it along corpus callosum the patient will talk to you with his left half and right half independently - each time unaware what his other half was talking about a moment earlier. I believe Zen emulates such split on a microscopic scale. We all go thru similar state when we conceive an original idea - if we do not write this down or formulate it immediately the discovery will loose its punch and sense of depth. The best documented and most prominent example is Hegel's discovery of dialectical logic - after extensive writing about that at some point he honestly admits loosing the original concept. But if not for his western oververbosity we might not today use the term 'naive' set theory. Every serious mathematician touching foundations is today perfectly aware of the price we pay for formulating things. But I am unaware of any virtue of trying to clap with one hand;-) Regards, -Andrzej

1 0

haskell for windows mobile 6 on Intel PXA270 processor
by Constantine Potapov 22 Dec '08

22 Dec '08

Hi All. I'm looking for haskell interpreter/compiler for windows mobile 6 (I have an Asus A696 PDA with intel PXA270 processor). Somebody head about any haskell implementation for this platform? -- With best regards, Constantine Potapov.

1 0

Re: [Haskell] ANN: HLint 1.0
by Neil Mitchell 22 Dec '08

22 Dec '08

Hi Lennart, > It would be nice if HLint didn't suggest things that it will object to > in the next round. > Like > > LLVM/Core/CodeGen.hs:176:1: Eta reduce > Found: > applyArgs f g = apArgs 0 f g > Why not: > applyArgs f = apArgs 0 f The idea is to specify things one step at a time, so the user learns from the hints, as well as fixing them. However, in the case of eta reduction, doing several reductions at a time does make sense. I'll add it to the todo list. Thanks Neil

1 0

Removing/Uninstalling cabal packages ?
by Laurent Giroud 22 Dec '08

22 Dec '08

Hi, I have been doing a few experiments with cabal packages lately and I wish to uninstall these to return to a cleaner package base. However, there doesn't seem to be a "cabal uninstall" command and reading the documentation at http://www.haskell.org/ghc/docs/latest/html/Cabal/index.html does not tell much about how to do that manually. Note that what I wish to do is both delete the installed files and the cabal database so that the uninstalled packages does not show anymore when running "cabal list --installed". Would anyone know how to do that ? Thanks in advance. Regards, Laurent -- Laurent Giroud ml(a)niaow.com

3 2

RE: [Haskell-cafe] Trouble with FFI on Windows2000...
by Serge LE HUITOUZE 22 Dec '08

22 Dec '08

Hi Kyra (I tried to email you directly, but, for some reason, my mail bounced...), Hi Haskellers, > > dlltool -d cproj1.def -l libcproj1.a > > ghc --make testFFI_2.hs -optl-lcproj1 -optl-L. > > [...] > > This leads to an error box with message: > > "DLL (null) not found on specified path" > > dlltool -d cproj1.def -l libcproj1.a -D cproj1.dll Thanks very much for this answer, short and effective (an ideal one)! Just an additional question, just for curiosity: Does anyone know why this "-D" option doesn't appear in the following two pages? - http://www.haskell.org/haskellwiki/GHC/Using_the_FFI#Setting_up_your_build_… - http://www.haskell.org/haskellwiki/GHC:FAQ#GHC_on_Windows (parag. 1.4.3.4) Is this because the behavior I got is Win2K specific (hence very few people run into in 2008)? Thanks again Kyra for your precious help. --Serge

1 0

Yampa vs. Reactive
by Tony Hannan 22 Dec '08

22 Dec '08

Hello, Can someone describe the advantages and disadvantages of the Yampa library versus the Reactive library for functional reactive programming, or point me to a link. Thanks, Tony P.S. It is hard to google for Yampa and Reactive together because "reactive" as in "function reactive programming" always appears with Yampa

5 17

ANN: llvm-0.4.0.1
by Lennart Augustsson 22 Dec '08

22 Dec '08

We have now released version 0.4.0.1 of the Haskell LLVM bindings. (This release that is quite incompatible with the old 0.0.2 release.) LLVM is a virtual machine and the bindings allow you to generate code for this virtual machine. This code can then be executed by a JIT or written to a file for further processing by the LLVM tools. The LLVM bindings has two layers. You can either use the low level bindings that is just the same as the C bindings for the LLVM. This level is quite unsafe as there is very little type checking. The recommended way is a high level binding (somewhat less complete) which eliminates many errors by leveraging the Haskell type system. A simple example, generating code for a function that adds two numbers and then calling it from Haskell: import Data.Int import LLVM.Core import LLVM.ExecutionEngine cgplus :: CodeGenModule (Function (Int32 -> Int32 -> IO Int32)) cgplus = createFunction InternalLinkage $ \ x y -> do r <- add x y ret r main = do ioplus <- simpleFunction cgplus let plus = unsafePurify ioplus print $ plus 20 22 Enjoy! Bryan O'Sullivan Lennart Augustsson

1 0