June 2007 - Haskell-Cafe

Re: [Haskell-cafe] Practical Haskell question.
by Jon Cast 25 Jun '07

25 Jun '07

On Monday 25 June 2007, Michael T. Richter wrote: > On Mon, 2007-25-06 at 01:05 -0500, Jon Cast wrote: > > What exactly are you trying to do? > > I'm trying to model a transactional system which could fail > mid-transaction without an easy rollback on conditions which could be > checked in advance. <snip> > Now if I could just check permissions on each operation > the monad emerging is clearer. Each call in the do block to C, then D > and finally E would be individually checked for permissions in the Monad > before the next step could progress. I can do this. But if I want to > change the model to pre-check the privileges (because of performance > issues or because of resource allocation issues or whatever) I see no > easy way to do the same thing. This intrigues me. Why? There's no easy or simple way to tell what operations you're going to execute in advance. Take a concrete example: do s <- readFile "/etc/ourProg" let Right xn = flip parse s $ many $ fileName flip mapM xn $ \ filename -> readFile xn Now, you can't possible pre-check this for permissions, because the argument to readFile in the fourth line is drawn from an external file; you would have to check for permission to read every file in the world. If your particular permissions issues are indeed independent of the arguments to performB, you can use an arrow: data MyArrow alpha beta = MyArrow (IO Bool) (alpha -> IO beta) instance Arrow MyArrow where arr f = MyArrow (return True) (return . f) MyArrow c0 a0 >>> MyArrow c1 a1 = MyArrow (liftM2 (&&) c0 c1) (\ x -> a0 x >>= a1) Then you can check the pre-conditions beforehand. (If you want to see why using an arrow helps, try implementing an instance of the class class Arrow a => ArrowApply a where apply :: a (a alpha beta, alpha) beta for MyArrow. Can't be done, but apply is built in to monads). Sincerely, Jonathan Cast http://sourceforge.net/projects/fid-core http://sourceforge.net/projects/fid-emacs

1 0

Practical Haskell question.
by Michael T. Richter 25 Jun '07

25 Jun '07

Now I've got a situation I can't figure out how to resolve. I want to have a set of actions which are executed sequentially, but which, before I even start to execute the first one, have been inspected for legality and/or plausibility. Consider this kind of sequence: do x <- performActionA y <- performActionB z <- performActionC return $ calculateStuff x y z Now obviously this is going to be in a monad of some kind. Were this a regular, run-of-the-mill program I'd just use the IO monad. But what I want to do instead is, before executing any of the perform* functions, check that the actions desired are actually permitted (or possible) given a set of circumstances. For example let's say it's a permissions issue and performActionB can only be done if I'm root. If I'm not root I don't want performActionA done because I can't complete the transaction. (Maybe ActionA is non-reversible, say.) Or let's say this is code that's accessing databases on the network. If the network link to C can't be established, I don't want to screw around with A and B's links at all -- it's too expensive, too time-consuming or whatever. Were I programming this in C, C++, Python, Ruby, etc. I could do this in my sleep. Functions are addresses (C/C++) or objects with an ID (Python/Ruby) so it's possible to take them and do some kind of check based on identities before executing things (although the scaffolding around this would be nontrivial in any of these languages except, possibly, Ruby). Functions in Haskell don't have this property, however, so I can't figure out what I'd do to perform similar work. I'm sure there's a way to do it, but I just can't see it. -- Michael T. Richter <ttmrichter(a)gmail.com> (GoogleTalk: ttmrichter(a)gmail.com) I'm not schooled in the science of human factors, but I suspect surprise is not an element of a robust user interface. (Chip Rosenthal)

14 23

FFI and Excel VBA
by Lewis-Sandy, Darrell 25 Jun '07

25 Jun '07

I am new to using the Haskell FFI, and have been trying to implement the example in section 11.6 of the GHC user's guide. I have finally gotten to the point where my dll compiles (there is a missing space in the mainDll.h code at line 4:12) using GHC 6.6.1, and used the declare statement to expose the "adder" function in Excel VBA. My VBA Code is given below: Private Declare Function adder Lib " adder.dll" Alias "adder@8" (ByVal x As Integer, ByVal y As Integer) As Integer Private Sub test() Debug.Print adder(1, 2) End Sub My problem is this: The function works fine (the immediate window displays 3), but when I terminate Excel, I get an application error ("The instruction at ... referenced memory at ... . The memory could not be read."). Does anyone else have any experience calling Haskell from VBA that might be relevant?

4 3

SYB with class, simplified
by Benja Fallenstein 25 Jun '07

25 Jun '07

Hi all, The "scrap your boilerplate with class" sytstem [1] has two big advantages over the plain SYB system from Data.Generics, IMHO: One, it lets you declare an 'open' generic function as a type class, to which new cases can be added by adding new instances (emphasized in the paper); and two, it lets you write recursive functions that require other type class constraints in addition to Data (not emphasized in the paper, but something I've frequently found myself wanting with Data.Generics). [1] http://homepages.cwi.nl/~ralf/syb3/ However, when trying to convert the codebase I'm working on to SYB-with-class, I've found that the type proxies and explicit dictionaries used to simulate type class abstraction over type classes are... annoying. Today, I've hit on an alternative approach to implementing SYB-with-class (YAGS, yet another generics scheme...), with less boilerplate per generic function. The approach may or may not be new (I haven't studied *all* of the generics proposals out there yet); in any case, it shares the use of type-level functions with Smash Your Boilerplate, and it uses the same underlying gfoldl operator as SYB, but implements it in a quite different way. I believe that the equivalent of everywhere, mkT and friends can be implemented as type-level functions in this framework, but I haven't actually tried it yet. This mail is a literate script demonstrating the approach. I'm hoping to get some feedback on the idea. :) On to the code: > {-# OPTIONS_GHC -fglasgow-exts -fallow-overlapping-instances > -fallow-undecidable-instances #-} Yup, we need it all... I'll start with three example generic functions. * 'size' calculates the number of constructors in a term, except for lists, for which it returns one plus sum of the element sizes. * 'inc' increases all Ints in a term by one. * 'prints' prints out each subterm of a term on its own line, except for strings, for which it prints the string, but not its subterms. Thus, the following code: > test = ("Hello", 7::Int, [2,3::Int]) > main = do print (size test); print (inc test) > putStrLn ""; prints test; return () prints this: ---------------------------------------------------------------------- 11 ("Hello",8,[3,4]) ("Hello",7,[2,3]) "Hello" 7 [2,3] 2 [3] 3 [] ---------------------------------------------------------------------- Here is the 'size' function: > class Size a where size :: a -> Int > > data SizeF = SizeF > instance Size a => Apply SizeF a Int where apply _ = size > > instance Size a => Size [a] where size xs = 1 + sum (map size xs) > instance Apply (GMapQ SizeF) a [Int] => Size a where > size x = 1 + sum (gmapQ SizeF x) The constraint (Apply f x r) means that 'f' is a type-level function that, when applied to 'x,' returns 'r': > class Apply f x r | f x -> r where apply :: f -> x -> r Here is the 'inc' function: > class Inc a where inc :: a -> a > > data IncF = IncF > instance Inc a => Apply IncF a a where apply _ = inc > > instance Inc Int where inc = (+1) > instance Apply (GMapT IncF) a a => Inc a where inc = gmapT IncF And here is the 'prints' function; for illustration, the implementation is in a slightly different style, which does without the declaration of a new type class: > data PrintsF = PrintsF; prints x = apply PrintsF x > instance Apply PrintsF String (IO String) where > apply _ x = print x >> return x > instance (Show a, Apply (GMapM PrintsF) a (IO a)) => > Apply PrintsF a (IO a) where > apply f x = print x >> gmapM f x Note the 'Show' constraint: 'prints' can only be applied to values all of whose subterms implement 'Show.' This is the kind of constraint you can't have with the standard, not-with-class SYB code. So much for the demo code; now, onwards to the actual library. The core consists of the following three type classes: > class Constr x f where constr :: x -> a -> f a > class Param x p f where param :: x -> f (p -> a) -> p -> f a > > class GFoldl x a f where gfoldl :: x -> a -> f a Together, these classes form the equivalent of the standard SYB's 'gfoldl' method. (I'm ignoring the rest of the Data class at this time, but I believe that it could be implemented in a similar fashion.) * 'Constr' and 'Param' correspond to the first and second argument of the standard SYB's gfoldl. * The parameter 'x' specifies the type of fold to perform (GMapQ, GMapT and GMapM in the present module). * We give an instance 'Constr' and 'Param' for each type of fold. We give an instance of 'GFoldl' for each type we want to fold over. Here are the instances of GFoldl: > instance Constr x f => GFoldl x () f where gfoldl = constr > instance Constr x f => GFoldl x Char f where gfoldl = constr > instance Constr x f => GFoldl x Int f where gfoldl = constr > > instance (Constr x f, Param x a f, Param x [a] f) => GFoldl x [a] f where > gfoldl x [] = constr x [] > gfoldl x (y:ys) = constr x (:) `p` y `p` ys where > p a b = param x a b > > instance (Constr x f, Param x a f, Param x b f, Param x c f) => > GFoldl x (a,b,c) f where > gfoldl x (a,b,c) = constr x (,,) `p` a `p` b `p` c where > p a b = param x a b What remains is the code for GMapQ, GMapT and GMapM: > newtype GMapQ f = GMapQ f; gmapQ f = apply (GMapQ f) > > newtype K a b = K { fromK :: a } > > instance GFoldl (GMapQ f) a (K [r]) => Apply (GMapQ f) a [r] where > apply (GMapQ f) x = reverse $ fromK $ gfoldl (GMapQ f) x > > instance Constr (GMapQ f) (K [r]) where constr _ _ = K [] > instance Apply f a r => Param (GMapQ f) a (K [r]) where > param (GMapQ f) (K xs) x = K (apply f x : xs) > newtype GMapT f = GMapT f; gmapT f = apply (GMapT f) > > newtype I a = I { fromI :: a } > > instance GFoldl (GMapT f) a I => Apply (GMapT f) a a where > apply (GMapT f) x = fromI $ gfoldl (GMapT f) x > > instance Constr (GMapT f) I where constr _ = I > instance Apply f a a => Param (GMapT f) a I where > param (GMapT f) (I x) y = I (x (apply f y)) > newtype GMapM f = GMapM f; gmapM f = apply (GMapM f) > > instance (Monad m, GFoldl (GMapM f) a m) => Apply (GMapM f) a (m a) where > apply (GMapM f) x = gfoldl (GMapM f) x > > instance Monad m => Constr (GMapM f) m where constr _ = return > instance (Monad m, Apply f a (m a)) => Param (GMapM f) a m where > param (GMapM f) m x = do fn <- m; arg <- apply f x; return (fn arg) That ends the example. Comments would be appreciated! :-) Thanks, - Benja

1 0

Constructing a datatype given just its constructor as a string?
by Hugh Perkins 25 Jun '07

25 Jun '07

Trying to create a datatype/constructor given just its constructor as a string, something like: mkConstr :: String -> Constr parseData :: (Data a) => String -> a ***without knowing in advance anything about the datatype apart from the string contents*** So, not something like: parseData "Just 3" :: [Maybe] but simply: parseData "Just 3" (This is linked to the other post: the goal is to be able to deserialize xml without needing to know in advance the data type we are deserializing)

5 5

Re: [Haskell-cafe] Installing HOpenGL
by Thomas Schilling 25 Jun '07

25 Jun '07

No, ffi is no extra package. It would be helpful if you posted you configure output though. BTW, please CC the list when replying. The easiest way should be to use the "reply all" feature of your mail client. Thanks, Thomas On 24 jun 2007, at 15.24, Jimmy Miller wrote: > That's all I know right now. Is there another package I need to > install for the C FFI? Also, is there any problem in Setup.hs saying > that sh was not found? > > On 6/24/07, Thomas Schilling <nominolo(a)googlemail.com> wrote: >> That looks like a missing C macro definition. It should probably >> expand to either stdcall or ccall. A bit more information would be >> helpful. >> >> >> On 24 jun 2007, at 13.11, Jimmy Miller wrote: >> >> > I'm trying to install HOpenGL 2.1, with ghc 6.6 on Windows XP. >> > runghc Setup.hs configure seems to work fine. It gives messages >> about >> > things missing, like sh, but there are no errors. >> > >> > When I run runghc Setup.hs build, I get the following error: >> > >> > [ 4 of 81] Compiling Graphics.Rendering.OpenGL.GL.Extensions >> > ( Graphics/Renderin >> > g/OpenGL/GL/Extensions.hs, dist\build/Graphics/Rendering/OpenGL/GL/ >> > Extensions.o >> > ) >> > >> > Graphics/Rendering/OpenGL/GL/Extensions.hs:46:15: >> > parse error on input `CALLCONV' >> > >> > Does anyone know what might be going wrong? >> > >> > -- >> > <a href="http://www.spreadfirefox.com/? >> > q=affiliates&id=153516&t=1"><textarea >> > rows="3" cols="40">Get Firefox!</a> >> > _______________________________________________ >> > Haskell-Cafe mailing list >> > Haskell-Cafe(a)haskell.org >> > http://www.haskell.org/mailman/listinfo/haskell-cafe >> >> > > > -- > <a href="http://www.spreadfirefox.com/? > q=affiliates&id=153516&t=1"><textarea > rows="3" cols="40">Get Firefox!</a>

2 11

Re: copy-on-write monad?
by oleg＠pobox.com 24 Jun '07

24 Jun '07

Greg Meredith wrote: > First, has anyone worked out a monadic > approach to copy-on-write? (And, Is there any analysis of perf > characteristics of said monadic schemes?) If you use Zippers (Huet's or generic ones) with functional updates, copy-on-write comes out automatically and by default. This is explained in http://okmij.org/ftp/Computation/Continuations.html#zipper and, in a more readable form, in a recent paper http://okmij.org/ftp/papers/context-OS.pdf The web page also contains the complete code.

2 1

Installing HOpenGL
by Jimmy Miller 24 Jun '07

24 Jun '07

I'm trying to install HOpenGL 2.1, with ghc 6.6 on Windows XP. runghc Setup.hs configure seems to work fine. It gives messages about things missing, like sh, but there are no errors. When I run runghc Setup.hs build, I get the following error: [ 4 of 81] Compiling Graphics.Rendering.OpenGL.GL.Extensions ( Graphics/Renderin g/OpenGL/GL/Extensions.hs, dist\build/Graphics/Rendering/OpenGL/GL/Extensions.o ) Graphics/Rendering/OpenGL/GL/Extensions.hs:46:15: parse error on input `CALLCONV' Does anyone know what might be going wrong? -- <a href="http://www.spreadfirefox.com/?q=affiliates&id=153516&t=1"><textarea rows="3" cols="40">Get Firefox!</a>

2 1

stupid operator question
by Michael Vanier 24 Jun '07

24 Jun '07

I noticed that both the Data.Array library and the Data.Map library use the (!) operator for different purposes. How would it be possible to import both libraries usefully in a single module? I guess what I'm really asking is: how do I qualify operator names? Mike

2 1

Re: [Haskell-cafe] avoiding command window with wxHaskell on Windows?
by Jens Fisseler 23 Jun '07

23 Jun '07

On Sat, 23 Jun 2007, Dean Herington wrote: > But if I double-click the .exe file, I get a command window that hangs around > (but doesn't appear to do anything, fortunately) until the program terminates. > How can I avoid this command window? With gtk2hs, using "-optl-mwindows" as a command line option for GHC lets me get rid of this window. Perhaps it will do the same for wxHaskell? Regards, Jens

1 0