- Haskell-Cafe - Haskell.org

More STUArray questions
by Martin Percossi 15 Mar '06

15 Mar '06

Hello, I am trying to write a haskell-native matrix library (providing a similar set of functionality as gsl's matrix/eigenvector routines). I have had a look at other attempts at matrix libraries, and have found Hal Daume's and Alberto Ruiz's libraries to offer a good amount of functionality. There are two major reasons for wanting to write my own library: 1. Haskell-nativeness: I have had some issues compiling and linking with gsl libraries on 64-bit platforms. Also, it would be quite interesting to gauge haskell's effectiveness as a scientific computing platform, in particular making use unboxed arrays as representation. 2. Use of monads: in the afore-mentioned libraries, monads are either ignored, or are restricted to the IO monad. I am taking a different approach, starting my code from the ST monad, and eventually this will be generalized to work with the IO monad. I think the ST monad is a good monad to be able to perform computations on matrices that update them, allowing efficient, in-place algorithms, but it also provides the benifit of not being a one-way street like the IO monad is. Being a relative newcomer to haskell, I would be interested to hear any thoughts as to whether this is a good/bad idea. Now to my question: I would like to represent a matrix as a wrapper around a block, which in turn is just an unboxed array. Here are the definitions for a matrix in ST and outside of a monad, respectively: type MBlock s = STUArray s Int Double data MMatrix s = MMatrix Int Int (MBlock s) type Block = UArray Int Double data Matrix = Matrix Int Int Block Now, I have started by providing some fairly low-level routines in the ST monad. My problem is depressingly simple: I would like to retrieve a matrix from the ST monad for use outside of it (in particular, to pretty-print it). Now, this is easy for STUArrays (just use runSTUArray), but I'm not sure of how to do it for a type that *encloses* an STUArray. For example, here's a simple test: runMatrix = do _A <- newListMatrix [[2, 0, 0], [0, 2, 0], [0, 0, 2]] _B <- newListMatrix [[1, 2, 1], [0, 1, 1], [0, 0, 1]] _C <- matMul _A _B return $ getBlock _C main = show $ runSTUArray runMatrix Obviously, what I get as a result of runSTUArray is a UArray, which is a pain because I'd then have to box it "by hand" (i.e. again specifying the number of rows and columns), instead of automatically, in order to have a Matrix again. So what I'd like is a runSTMatrix routine, which would possibly have signature: runSTMatrix :: ST s (Matrix s) -> Matrix which would have semantics analogous to runSTUArray. Does anyone have any idea of how I can write runSTMatrix? Many thanks in advance, Martin

5 10

fmap for lists of lists of lists of ...
by Greg Buchholz 15 Mar '06

15 Mar '06

Is it possible to make a typeclass like Functor, that has a function (say "f_map"), which would work for the infinite hierarchy of types: ([],[[]],[[[]]],...)? Does that make sense? This doesn't work... >class Funct f where > f_map :: (a -> b) -> f a -> f b > >instance Funct [] where > f_map = map > >instance Funct a => Funct [a] where > f_map f = map (f_map f) > >main = print $ fmap (+1) [[[1,2,3]]] ...but maybe it gets the idea across? Thanks, Greg Buchholz

1 0

GUI-Woes
by Daniel Fischer 15 Mar '06

15 Mar '06

Hello All, how would I get myself a working (and easy to use) GUI-library? Fudgets looks good, but I couldn't get it to build. I downloaded the snapshot from 2005, how to compile says make ghc-all; make ghc-install, but it doesn't work. First, in several scripts/makefiles, the ghc version 4.06 was hardwired, after some editing make complained about not finding lmlmk, whatever that is. If that's all that I would need to build it successfully, where would I get it? I also tried FranTk, doesn't look bad either, but has ghc -version 4.04 hardwired, imports IOExts, which I think would have to be replaced with GHC.Exts, and Addr, which I have no clue what it should be now, anyway, it seems pre-hierarchical-libraries, which would mean a lot of editing the source-files. So, since Fudgets does work in principle, - which version should I use and where do I get it? - what else would I have to download to get it to build? - in case it isn't straightforward, how to build it? I have a linux machine with ghc-6.4.1 and hugs, no HBC, NHC, JHC,..., no GTk2hs (would I need that?). Thanks, Daniel -- "In My Egotistical Opinion, most people's C programs should be indented six feet downward and covered with dirt." -- Blair P. Houghton

4 7

Re: [Haskell] Trying to get a Composite design pattern to work
by Jared Updike 15 Mar '06

15 Mar '06

(Moved to haskell-cafe) > Actually, I'm trying to avoid library functions, so I can learn the > language and the functional way of thinking. How would one implement > the concatMap function? The Haskell 98 report gives possible implementations of standard functions: http://haskell.org/onlinereport/standard-prelude.html e.g. foldr :: (a -> b -> b) -> b -> [a] -> b foldr f z [] = z foldr f z (x:xs) = f x (foldr f z xs) map :: (a -> b) -> [a] -> [b] map f [] = [] map f (x:xs) = f x : map f xs concat :: [[a]] -> [a] concat xss = foldr (++) [] xss concatMap :: (a -> [b]) -> [a] -> [b] concatMap f = concat . map f etc. The book Haskell School of Expression (by Paul Hudak) goes over using recursion as a control structure and then shows you how to replace your simple recursion patterns with library functions once you start to see these general patterns. As you start to understand the functional style of iterating over lists, you should begin to see when to use the library functions, and how it will save you time and be more simple and clear than using recursion directly for every function. If you want a place to start, I would challenge you to expand the concatMap definition above with the definitions of concat and map, and then plug in the definition of foldr and see if you can make your own concatMap function. Maybe that will help you understand things better. Hope that helps, Jared. -- http://www.updike.org/~jared/ reverse ")-:"

4 4

Tree with collections
by Henning Thielemann 13 Mar '06

13 Mar '06

I have a problem with defining an instance. > module CollTree where > import Data.List(intersperse) Say, I want to define a tree structure based on lists. I omit the data attached to the nodes here for simplicity. > data ListTree = ListNode [ListTree] > deriving Show This is simple enough for an automatically derived Show instance. Now I want to generalise that structure from lists to arbitrary types of collections. > data CollTree coll = CollNode (coll (CollTree coll)) This got too complicated for automatic derivation. So I have to define a Show instance for CollTree manually. But how to formulate the instance context? How to avoid undecidable instances? instance ??? => Show (CollTree coll) where show (CollNode xs) = "CollNode " ++ show xs Currently I see only one solution using a new class providing a special 'show' routine, which handles the structure of the collection, but must be assisted by a custom 'show' function for the collection members. > class CollShow coll where > collShow :: (a -> String) -> (coll a -> String) > instance CollShow coll => Show (CollTree coll) where > show (CollNode xs) = "CollNode " ++ collShow show xs Now I have to write 'show' related code for each collection type. This way I probably duplicate a lot of code that is already written for the Show instances of the collections. To be honest, I use a more special tree structure with even more special "collections" so this may be not really a problem. Here an instance for a custom list type: > newtype NewList a = NewList [a] > instance CollShow NewList where > collShow shw (NewList xs) = > "(NewList [" ++ > concat (intersperse ", " (map shw xs)) ++ "])" Is there a more straightforward way, preferrably Haskell98?

2 2

Re: UArray
by Bulat Ziganshin 13 Mar '06

13 Mar '06

Hello Frederik, Sunday, March 12, 2006, 9:39:13 PM, you wrote: FE> Why define separate instances when one could just define one FE> equivalent one: FE> instance (Ix ix, Eq e, IArray UArray e) => Eq (UArray ix e) FE> (==) = eqUArray FE> ??? afaik, this is just not H98-compatible and require to use even more type extensions than currently used in this module FE> The same question applies to the Ord instances in that file. FE> I can give an answer *why not* do this: it makes UArray even more FE> unusable. UArrays which are polymorphic in their element types cannot FE> be compared. Here is another patch, please apply: _I_ don't apply the patches, but you can propose patches by sending them to libs maillist. tomorrow Simon Marlow will read it and decide whether it need to be applied most UArrays users don't need polymorphism, i run into problems with arrays only when i wrote general library which should support every array type -- Best regards, Bulat mailto:Bulat.Ziganshin@gmail.com

1 0

Re: Can't uninstall GHC or prepare SOE installation on OS X
by Xavier Elizalde 12 Mar '06

12 Mar '06

Hi Wolfgang, For some reason, GHCi now keeps launching every time I launch Terminal.app, I can't even get a regular command line prompt now. I had to uninstall all the files for GHCi manually from an X11 Xterm to get a regular prompt. It still gets an error trying to launch GHCi, but then ends up with a regular prompt. I don't know why it started doing that, it wasn't doing that before.

2 2

Re: Can't uninstall GHC or prepare SOE installation on OS X
by Wolfgang Thaller 09 Mar '06

09 Mar '06

Xavier Elizalde wrote: > Hi. I sent this post to the haskell-cafe mailing list but haven't > gotten any replies. Ah well, I watch that list in digest mode only, easy to overlook things. > Basically, I can't get the Uninstall AppleScript for GHC 6.4.1to > run properly because it encounters an error. Oops. I *thought* I had tested that to death. Anyway, for manual removal: /usr/local/lib/ghc-6.4.1 (the big part; if disk space is all you're after, stop here) In /usr/local/bin/: ghc, ghc-6.4.1, ghc-pkg, ghc-pkg-6.4.1, ghci, ghci-6.4.1, ghcprof, hasktags, hp2ps, hsc2hs, runghc, runhaskell, stat2resid in /Libraries/Receipts: Some GHC-*.pkg file, I suspect now that it's called GHC-6.4.1.pkg /Library/Frameworks/GMP.framework (do not delete this unless you do not want to ever run any GHC-compiled program again until you re- install GHC). > I have the Haskell School of Expression book and want to try out > the examples in the book that make use of X11 for graphics, and I > don't know how to use GHC 6.4.1 for the SOE book examples. I > thought it would be easiest to just uninstall it then install the > SOE binaries for OS X. The SOE binaries use hugs, and should therefore be completely independent of GHC, no uninstalling required. > But if GHC 6.4.1 works fine for the SOE book examples and is better > than the SOE install, I'd like to use that, although I don't know > how to get it to work with X11. Do you have any idea on what I > should do? Start your programs from an xterm. If you want to use GHCi, set export DYLD_LIBRARY_PATH=/usr/X11R6/lib/ before starting ghci in the xterm. > > I have The Haskell School of Expression book and the > instructions for setting up and running the SOE software seem to have > a different configuration compared to GHC 6.4.1. Namely, it's meant > to use X11 and locations for files are different. What files? I don't have the book, so all I can tell you is that GHC includes a library module by the name of Graphics.SOE. > Or does the GHC already > have GraphicsLIb as part of it? If so, shouldn't it launch an X11 > Xterm terminal rather than the standard OS X Terminal? Shouldn't it launch...? Are you referring to the AppleScript for launching GHCi? If so, then no, for two reasons: a) X11 is an optional install on Mac OS X, so many people don't have it. b) The standard Terminal is a lot more convenient to use than an xterm, and most people don't need/want to use X11 for their Haskell programs. So good luck with SOE, if there are any more problems, ask me. Cheers, Wolfgang

3 2

Question regarding let clauses
by Martin Percossi 09 Mar '06

09 Mar '06

Hello, the following code doesn't compile <snip> module Matrix where import Control.Monad.ST import Data.Array.ST import Data.Array.Unboxed type Block s = STUArray s Int Double data MMatrix s = MMatrix Int Int (Block s) newMatrix_ :: Int -> Int -> ST s (MMatrix s) newMatrix_ m n = do b <- newArray_ (1, m*n) return (MMatrix m n b) runMatrix = do _A <- newMatrix_ 3 3 _B <- newMatrix_ 3 3 matMul _A _B return "Success" main = show $ runST runMatrix matMul :: MMatrix s -> MMatrix s -> ST s (MMatrix s) --matMul a b = do let foo = 2*5 --return a matMul a b = do { let foo = 2*5; return a } </snip> under ghc 6.4.1, yielding the error message: question.hs:25:41: parse error on input `<-' Failed, modules loaded: none. The offending is line the one containing "let foo = 2*5", which is a little test I've done of let-clauses. Now, suppose instead that for the last function, matMul, I replace the version that's commented out. No errors! Could someone enlighten me as to why? I'm a bit confused, as I thought the two forms are equivalent save for formatting... This is on the back of a email discussion that I was reading about let-clauses, in which someone declared that they where better than where clauses for monadic code. If anyone could comment on this, I'd appreciate it as well. Many thanks in advance, Martin

4 3

Haddock Problems
by Daniel Fischer 09 Mar '06

09 Mar '06

Hi all, I've just installed haddock-0.7, nice, but... > haddock -o h7doc -h -D h7doc/fusi.haddock --use-package=base Verwaltung.hs Teams.hs Stats.hs Match.hs Main.hs Liga.hs Item.hs Helpers.hs Datum.hs Warning: Helpers: could not find link destinations for: GHC.Base.Int GHC.Base.String GHC.Show.Show Data.Maybe.Maybe GHC.Base.Eq GHC.Base.Bool and so on. Same problem, if I use "-i/usr/.../base.haddock" instead of --use-package=base. This doesn't happen with haddock-0.6, all are properly linked to Prelude.html then. Now what can I do about that? BTW, in the ghc-docs (created with haddock-0.6), in most html-pages, Int &c are linked to GHC.Base, GHC.Num... which don't exist, how would I manage to have those linked to Prelude.html? And when I tried to re-make install-docs (for ghc-6.2.2, not to put the 6.4.1-docs into peril), I got /bin/sh: line 1: t: command not found make[2]: [html/haddock.css] Fehler 127 (ignoriert) /usr/bin/install -c -m 644 html/* /usr/local/share/ghc-6.2.2/html/libraries/base /usr/bin/install: Aufruf von stat für »html/*« nicht möglich: Datei oder Verzeichnis nicht gefunden make[2]: *** [install-docs] Fehler 1 make[1]: *** [install-docs] Fehler 1 make[1]: Leaving directory `/usr/local/lib/ghc-6.2.2/libraries' make: *** [install-docs] Fehler 1 linux:/usr/local/lib/ghc-6.2.2 # Apparently had a problem with SRC_HADDOCK_OPTS += -t "Haskell Hierarchical Libraries..." How can I fix all that? Thanks, Daniel -- "In My Egotistical Opinion, most people's C programs should be indented six feet downward and covered with dirt." -- Blair P. Houghton

2 5