- Haskell-Cafe - Haskell.org

syntactic anti-unification of TH terms --- best practise?
by Martin Hofmann 18 Jul '08

18 Jul '08

I am implementing syntactic anti-unification for TH terms (Exp, Pat, Clause, ...). For example anti-unifying the clauses tail (1:xs) = xs tail (1:2:xs) = (2:xs) would yield tail (1:x1) = x1 whereas the anti-instance of last (1:[]) = 1 last (1:xs) = last xs is last (1:x1) = x2 So different subterms are always replaced with the same variable if anti-unified twice. So far, so good, there is nothing complicated. I am using a State Monad to keep track of my variables and defined a type class for anti-unification and implemented instances for Exp, Pat, ... type AU a = State (Data.Map [a] a) a class Antiunifieable t where antiunify :: (Ord t) => [t] -> t antiunify = (\t -> evalState (aunify t) M.empty) aunify :: (Ord t) => [t] -> AU t Problems arise when I come to clauses, defined in TH as Clause [Pat] Body [Dec] where Body is for example NormalB Exp Since a term can occur both as a pattern on the left-hand side of the equation and also as expression on the right-hand side, I need to keep track of this, too. So in fact, I have to anti-unify the patterns, 'translate' the resulting state (Data.Map [Pat] Pat) to (Data.Map [Body] Body) and pass it to the state transformer for anti-unifying the bodies. Similarly, from Body to Exp. This seems a bit ugly to me, but I have no idea how to solve it more elegantly. I thought about introducing some kind of wrapper type like data Term = TC Clause | TB Body | TP Pat | TE Exp and implement an instance for Eq and Antiunifieable for it. But again, this is not better either and I run into a lot of bookkeeping when processing it. Is there some kind of best practise dealing with such a problem? I am not sure if MonadTransformers would help here, because it is always the same monad but with different parameter. Thanks a lot, Martin

3 3

Space leaks
by Peter Gavin 18 Jul '08

18 Jul '08

Hello everyone, I have this piece of code I've been working on, and I've been stuck on tracking down a space leak in it for some time now. The code is essentially a tight loop that updates a rather largish data structure with embedded functions that are called by the driver loop. The code doesn't accumulate any data as the loop runs (at least deliberately), so I would expect the memory profile to be flat. Unfortunately, the profile is a wedge :) I've added bangs and `seq` literally everywhere, and it looks (to me at least) like there's nothing left to be lazily evaluated anywhere. I've used retainer profiling, and the functions that are leaking space according to the profiler output are strict throughout. I'm really pulling my hair out over this, but I'm reluctant to publish the code just yet because I'm planning on using it (eventually) for my thesis. It seems like I've just about run out of options, though. Does anyone have any advice on tracking down space leaks? I don't want to accuse GHC of having a bug just yet, but has GHC had space leak bugs in the past? Thanks in advance, Peter Gavin <gavin(a)cs.fsu.edu>

4 5

FFI and struct arguments
by Felipe Lessa 17 Jul '08

17 Jul '08

Hi, I tried googling and searching the haskellwiki about this but wasn't lucky enough. My question is: is there a way to send struct arguments to C functions via the FFI or do I need to create a C wrapper? I guess there isn't, and while I can live without it, I'd like to leave no doubt. Details: I have something like ==================== typedef struct vect { float x,y; } vect; void func(vect v); ===================== on the C side and ==================== -- Please disregard float /= Float, just an example :) data Vector = Vector Float Float instance Storable Vector where ... ==================== on the Haskell side, and I want to call func with Vector as argument. Now, Vector isn't a basic FFI type, although it implements Storable. So does that mean that I need to create something like ==================== void funcWrapper(vect *v) { func(*v); } ==================== and then allocate some temporary memory on the Haskell side to use func? Cheers! -- Felipe.

5 8

uvector and the stream interface
by stefan kersten 17 Jul '08

17 Jul '08

currently i'm working on stuff that looks something like this: 1 read soundfile from disk in blocks of N samples (IOCArray, hsndfile package) 2 convert to CArray with unsafeFreeze (simple O(1) cast, carray package) 3 perform FFT (CArray, fftw package) 4 convert to UArr (uvector package) 5 do some stuff with vectors i'd like to minimize copying, and since the FFT returns a new array anyway, the only copying conversion is the one between CArray and UArr in step 4. the approach i've been following is defining a `stream' conversion for CArray, hoping that subsequent carray accesses will be fused with uvector operations without the need for allocating a vector in between. does that make sense? could this be a general strategy for avoiding copying at the boundary between the various array libraries? thanks, <sk>

9 14

Re: [Haskell-cafe] Type level sets with GADTs, fundeps etc
by J.Burton＠brighton.ac.uk 17 Jul '08

17 Jul '08

At Tue, 15 Jul 2008 10:27:59 -0400, Jeff Polakow wrote: > > [1 <text/plain; US-ASCII (7bit)>] > > [2 <text/html; US-ASCII (7bit)>] > Hello, > > > data LSet t where > > Nil :: LSet Nil > > Ins :: (Member a t b > > , If b t (a ::: t) r) > > => L a -> LSet t -> LSet r > > > Try replacing both original occurrences of r, i.e. (untested) > > Ins :: (Member a t b, If b t (a ::: t) (LSet r)) => L a -> LSet t -> LSet r > Thanks. This sort of works, but shifts the problem to another context. Now it seems that I can't hide the extra type information in the existential types, which is what I want to do. In the function `insertChar' below I want the type LSetBox to be opaque (i.e. it will be called by users who don't need to know about the fancy types): data LSet t where Nil :: LSet Nil Ins :: (Member a t b , If b t (a ::: t) (LSet r)) => L a -> LSet t -> LSet r -- I have to supply a type for `insert' now and it must include the constraints insert :: (Member a t b , If b t (a ::: t) (LSet r)) => L a -> LSet t -> LSet r insert = Ins --insertChar (and the boxing) doesn't work insertChar :: Char -> LSetBox -> LSetBox insertChar c (LSetBox s) = case fromChar c of LBox t -> LSetBox (insert t s) The error: Could not deduce (If b t1 (a ::: t1) (LSet t), Member a t1 b) from the context () arising from a use of `insert' at /home/jim/sdf-bzr/dsel/TF/Set-July08.hs:54:25-34 Possible fix: add (If b t1 (a ::: t1) (LSet t), Member a t1 b) to the context of the constructor `LBox' or add an instance declaration for (If b t1 (a ::: t1) (LSet t)) In the first argument of `LSetBox', namely `(insert t s)' In the expression: LSetBox (insert t s) In a case alternative: LBox t -> LSetBox (insert t s) Failed, modules loaded: none. Jim > -Jeff > > --- > > This e-mail may contain confidential and/or privileged information. If you > are not the intended recipient (or have received this e-mail in error) > please notify the sender immediately and destroy this e-mail. Any > unauthorized copying, disclosure or distribution of the material in this > e-mail is strictly forbidden. > >

2 1

Searchig for modules
by fero 17 Jul '08

17 Jul '08

Hi I am learnig Haskell form SOE book, but I can't find out how to import SOE module. I use eclipsefp IDE and I get no error when I have SOE.hs in my project and in SimpleGraphics.hs I make "import SOE". But when trying to run it I get error: "Could not find module `SOE': Use -v to see a list of the files searched for." Where ghc/ghci search for modules? Is it possible put some modules somewhere that every Haskell project can find it (something as global classpath in java). And is it posible in eclipsefp to link some modules to project that this project will use them. As I wrote even when I have 2 modules in src in project I when trying to run one of then the others are not found. I can't find any site about searching for modules (in java there is a lot options, global/local classpath, directory in java runtime used only with this rt or directory in web server used in all apps run on this server...). Can you put a link? Thanks Fero -- View this message in context: http://www.nabble.com/Searchig-for-modules-tp18505770p18505770.html Sent from the Haskell - Haskell-Cafe mailing list archive at Nabble.com.

3 3

Fixed-Point Combinators
by Adrian Neumann 17 Jul '08

17 Jul '08

Hello, while studying for a exam I came across this little pearl: Y = (L L L L L L L L L L L L L L L L L L L L L L L L L L L L) where L = λabcdefghijklmnopqstuvwxyzr. (r (t h i s i s a f i x e d p o i n t c o m b i n a t o r)) posted by Cale Gibbard to this list. Now I'm wondering how exactly does one finde such awesome λ expressions? Is there some mathemagical background that lets one conjure such beasts? Adrian

4 3

Linker problems linking FFI import call in Windows
by PJ Durai 17 Jul '08

17 Jul '08

Greetings I am trying to import some functions from a windows DLL. I am getting strange errors. (This used to work with previous versions of GHC. 6.6 I think.). This is my import statement: foreign import stdcall "ace.h AdsConnect" adsConnect' :: CString -> Ptr CInt -> IO CInt This is the build output. C:\moonar>\bin\make.exe ghc --make -ffi main.hs -ffi adsapi.hs -optc -DWIN32 native.cpp -lace32 -L/hom e/pjdurai/dev/moonar -L/tmp/Advantage -o moonar.exe [1 of 2] Compiling AdsApi ( adsapi.hs, adsapi.o ) [2 of 2] Compiling Main ( main.hs, main.o ) Linking moonar.exe ... adsapi.o(.text+0x124):fake: undefined reference to `AdsConnect@8' adsapi.o(.text+0x260):fake: undefined reference to `AdsDisconnect@4' collect2: ld returned 1 exit status make: *** [main.exe] Error 1 It actually links without errors if I use 'ccall' instead of 'stdcall'. But the executable segfaults when the function is called. I had the same code working with 'stdcall' declaration a while ago. (With GCH 6.6 I think) I am using ghc-6.8.3 on Windows XP SP2. appreciate your time. thanks pj

3 4

ANN: Topkata
by c-bauer-olsbruecken＠t-online.de 17 Jul '08

17 Jul '08

Hi All, Topkata is a simple OpenGL Game written in Haskell. It's not very advanced. Goal so far is to guide a ball trough an labyrinth to the opposite corner. The web page shows an screenshot. It's only tested under Linux. http://home.arcor.de/chr_bauer/topkata.html Feedback&Patches are welcome. BTW, I had a lot of support in #haskell! Christoph Bauer

14 19

curl installation
by Thomas Fuhrmann 16 Jul '08

16 Jul '08

Hello, I'm trying to install the curl package 1.3.2.1 for haskell, I'm using Ubuntu and I've already installed curl-7.18.2. I can't get rid of the configuration error. According to the log entry the curl.h doesn't exist. Shell output: runghc Setup.hs configure Configuring curl-1.3.2.1... checking for gcc... gcc checking for C compiler default output file name... a.out checking whether the C compiler works... yes checking whether we are cross compiling... no checking for suffix of executables... checking for suffix of object files... o checking whether we are using the GNU C compiler... yes checking whether gcc accepts -g... yes checking for gcc option to accept ANSI C... none needed checking how to run the C preprocessor... gcc -E configure: error: curl libraries not found, so curl package cannot be built See `config.log' for more details. Part of Log entry: configure:2439: gcc -E conftest.c conftest.c:9:23: error: curl/curl.h: No such file or directory configure:2445: $? = 1 configure: failed program was: | /* confdefs.h. */ | | #define PACKAGE_NAME "Haskell curl package" | #define PACKAGE_TARNAME "curl" | #define PACKAGE_VERSION "1.1" | #define PACKAGE_STRING "Haskell curl package 1.1" | #define PACKAGE_BUGREPORT "dons(a)galois.com" | /* end confdefs.h. */ | #include <curl/curl.h> configure:2469: error: curl libraries not found, s o curl package cannot be built See `config.log' for more details. Any ideas?

3 3