- Haskell-Cafe - Haskell.org

typing hmatrix expressions
by Carter Schonwald 26 Apr '11

26 Apr '11

hello alberto, i've had some funny issues with using the cmap function, and i'd like understand what i'm doing wrong namely it wont correctly instantiate for vectors or matrices and I don't understand why. basically my question is: what is the correct type for the hmatrix expression mapMat f =liftMatrix $! cmap f I'd like to just give it a type like Storable a => (a->a)-> Matrix a -> Matrix a but i get an error message like Electrical.lhs:89:29: No instance for (Container Vector e0) arising from a use of `cmap' Possible fix: add an instance declaration for (Container Vector e0) In the second argument of `($!)', namely `cmap f' In the expression: liftMatrix $! cmap f In an equation for `mapMat': mapMat f = liftMatrix $! cmap f which i feel shouldn't be happening. is the type of cmap in some way too polymorphic, or am I just not turning on the right ghc type extensions for this to work, or is this something that would have been well typed prior to ghc 7? Or am i just not correctly understanding this type error and i'm making a simple mistake? thanks! -Carter

1 0

converting prefixes of CString <-> String
by Eric Stansifer 26 Apr '11

26 Apr '11

I have been reading Foreign.C.String but it does not seem to provide the functionality I was looking for. Let 'c2h' convert CStrings to Haskell Strings, and 'h2c' convert Haskell Strings to CStrings. (If I understand correctly, c2h . h2c === id, but h2c . c2h is not the identity on all inputs; or perhaps c2h is not defined for all CStrings. Probably this is all locale dependent.) I have an infinite Haskell String transferred byte-wise over a network; I would like to convert some prefix of the bytes received into a prefix of the String I started with. However, if I understand correctly, if "s" is a Haskell String it is not necessarily true that "c2h (take n (h2c s))" is a prefix of s for all n. So I have two questions: Given a CString of the form "cs = take n (h2c s)", how do I know whether "c2h cs" is a prefix of s or not? Is there a way to recognize whether a CString is "valid" as opposed to truncated in the middle of a code point, or is this impossible? Better yet, given a CString "cs = take n (h2c s)", is there a way to find the maximal prefix cs' of cs such that c2h cs' is a prefix of s? If s == s1 ++ s2, is it necessarily true that s == (c2h (h2c s1)) ++ (c2h (h2c s2))? If so, then I can perform my conversion a bit at a time, otherwise I'd need to start from the beginning of the cstring each time I receive additional data. In practice, I think my solution will come down to restricting my program to only using the lower 128 characters, but I'd like to know how to handle this problem in full generality. Thanks, Eric

2 3

Why does Cabal not choose the latest package?
by Emil Axelsson 26 Apr '11

26 Apr '11

Hello! I've had some of the usual problems with packages depending on multiple versions of another package. It seems the root of the hole problem was that I once attempted to run cabal install cabal-install This brought in a number of older packages (Cabal-1.8.0.6, containers-0.3.0.0, directory-1.0.1.2, filepath-1.1.0.4, process-1.0.1.5) to my local package DB, which led to inconsistencies when trying to install new packages. However, if I run cabal install cabal-install-0.10.2 things work just fine. So the problem is that it wants to install an older version of cabal-install. A session demonstrating the problem is found below. Why doesn't cabal always install the latest packages? Thanks! / Emil [emax@laptop:~]> ghc --version The Glorious Glasgow Haskell Compilation System, version 7.0.2 [emax@laptop:~]> cabal --version cabal-install version 0.10.2 using version 1.10.1.0 of the Cabal library [emax@laptop:~]> cabal install cabal-install Resolving dependencies... Downloading containers-0.3.0.0... ... [emax@laptop:~]> cabal install cabal-install-0.10.2 Resolving dependencies... Configuring cabal-install-0.10.2... Preprocessing executables for cabal-install-0.10.2... Building cabal-install-0.10.2... ...

2 1

-- Extension for "Pearls of Functional Algorithm Design" by Richard Bird, 2010, page 25 #Haskell
by caseyh＠istar.ca 26 Apr '11

26 Apr '11

-- Extension for "Pearls of Functional Algorithm Design" by Richard Bird, -- 2010, page 25 #Haskell -- This version assumes 3 disjoint ordered sets represented as lists. -- So either: x<y XOR x>y -- Since it uses lists it is no faster than the divide and conquer approach. -- I might try to convert this version to sorted arrays for -- O(log|X|+log|Y|+log|Z|) performance -- If I can figure out how to do it without suffering from "indexitis". smallest3'' :: Ord a => Int -> ([a], [a], [a]) -> a smallest3'' k ([],[],ts) = ts !! k smallest3'' k (zs,[],[]) = zs !! k smallest3'' k ([],ws,[]) = ws !! k smallest3'' k ([],ws,ts) = smallest'' k (ws,ts) smallest3'' k (zs,[],ts) = smallest'' k (zs,ts) smallest3'' k (zs,ws,[]) = smallest'' k (zs,ws) smallest3'' k (zs,ws,ts) = ~~~~case (a<b, b<c, a<c) of ~~~~~~(True, True, True) -> smallest3h'' k ((zs,p,ys),(ws,q),(ts,o,rs)) -- a<b<c ~~~~~~(True, False, True) -> smallest3h'' k ((zs,p,ys),(ts,o),(ws,q,us)) -- a<c<b ~~~~~~(False, True, True) -> smallest3h'' k ((ws,q,vs),(zs,p),(ts,o,rs)) -- b<a<c ~~~~~~(False, True, False) -> smallest3h'' k ((ws,q,vs),(ts,o),(zs,p,xs)) -- b<c<a ~~~~~~(True, False, False) -> smallest3h'' k ((ts,o,ss),(zs,p),(ws,q,us)) -- c<a<b ~~~~~~(False, False, False) -> smallest3h'' k ((ts,o,ss),(ws,q),(zs,p,xs)) -- c<b<a ~~~~where ~~~~~~p = (length zs) `div` 2 ~~~~~~q = (length ws) `div` 2 ~~~~~~o = (length ts) `div` 2 ~~~~~~(xs, a : ys) = splitAt p zs ~~~~~~(us, b : vs) = splitAt q ws ~~~~~~(rs, c : ss) = splitAt o ts ~~~~~~smallest3h'' k ((zs,p,ys),(ws,q),(ts,o,rs)) = ~~~~~~~~case (k<=p+q+o) of ~~~~~~~~~~(True) -> smallest3'' k (zs,ws,rs) ~~~~~~~~~~(False) -> smallest3'' (k-p-1) (ys,ws,ts) smallest'' :: Ord a => Int -> ([a], [a]) -> a smallest'' k ([],ws) = ws !! k smallest'' k (zs,[]) = zs !! k smallest'' k (zs,ws) = ~~~~case (a<b, k<=p+q) of ~~~~~~(True, True) -> smallest'' k (zs,us) ~~~~~~(True, False) -> smallest''(k-p-1) (ys,ws) ~~~~~~(False, True) -> smallest'' k (xs,ws) ~~~~~~(False, False)-> smallest''(k-q-1) (zs,vs) ~~~~where ~~~~~~p = (length zs) `div` 2 ~~~~~~q = (length ws) `div` 2 ~~~~~~(xs, a : ys) = splitAt p zs ~~~~~~(us, b : vs) = splitAt q ws

2 1

Haskell from SML - referrential Transparency?!
by Gregory Guthrie 26 Apr '11

26 Apr '11

I am pretty new to Haskell, so need some clarification. I am porting some code from SML, and getting a result that surprises me. I basically have some functions which work like this: f1 = fa fb fc test1 = do print "test1:" f1 But I ran a few tests, and got odd results - so I ran the same test function twice, and got different results - that was my surprise. I did this: f1 = fa fb fc f2 = fa fb fc test2 = do print "test1:" f1 f2 and I get different results from the two executions (f1,f2), even though they have exactly the same definition. Reversing their order, gives the exact same results (i.e. the results are still different, and in the same original order as f2;f1). Even doing (f1;f1) gives two different results. Seems to me that by referential transparency, I should always get the same result from the function(s). So, I added some Debug.trace to the argument functions which are used, and I get a trace from the first call(s), but none from the second one(s), although I do get the result from each. It is as if because of the laziness, it someone cached some of the intermediate results, so did not re-invoke the functions. Anyway, totally confused. I must be missing something significant here. Thanks for any clarification! (The original code is a bit long, so I did not include here...)

8 11

Data.Enumerator.Text.utf8 not constant memory?
by Skirmantas Kligys 26 Apr '11

26 Apr '11

Hi, I am learning iteratees, and as a starter project I wanted to use expat- enumerator to parse a 2 gigabyte XML file. I expected to be able to do what SAX does in Java, i.e. to avoid loading the whole 2 gigabytes into memory. For warm-up, I wrote an iteratee to count lines in the file, and it does load the whole file into memory! After profiling, I see that the problem was Data.Enumerator.Text.utf8, it allocates up to 60 megabytes when run on a 40 megabyte test file. Any suggestions how to fix Text.utf8, or what people do for parsing UTF-8 encoded text files with iteratees? Thanks! Here is my code and profiling results: http://i.imgur.com/XEI1v.png http://hpaste.org/46037/counting_lines_with_iteratees http://hpaste.org/46038/counting_lines_with_iteratees

3 5

errors while installing yesod 0.8
by Michael Litchard 26 Apr '11

26 Apr '11

Trying to install yesod 0.8 breaks when it's time to install monad-control. Google wasn't very helpful, nor was the error message I received mlitchard@apotheosis:~/monad-control$ cabal install Resolving dependencies... Configuring monad-control-0.2.0.1... cabal: Error: some packages failed to install: monad-control-0.2.0.1 failed during the configure step. The exception was: ExitFailure 11 note: I've been trying to use the cab command to manage my packages, I get the same error as above when I use cab instead of cabal.I mention this just in case there is some unforseen problem having to do with cab/cabal interaction. Has anyone experienced this problem, or know what I can do to get more useful error messages that might reveal the cause of the breakage?

5 20

Storing failing testcases for QuickCheck
by Sönke Hahn 25 Apr '11

25 Apr '11

Hi! I would like to have a library that would allow to use QuickCheck in the normal manner, but it would save test data for failing properties on the filesystem (maybe using the shiny new acid-state?). On consecutive test runs, the saved test data would be used first (before generating new arbitrary test data). I think, this would come in quite handy when doing test driven development. Any comments on this? Does something like this already exist? Cheers, Sönke

3 3

Trouble with currying and uncurrying...
by Angel de Vicente 25 Apr '11

25 Apr '11

Hi, I'm reading "The Craft of Functional Programming" and I found something I don't understand in page 185. It says: "Suppose first that we want to write a curried version of a function g, which is itself uncurried and of type (a,b) -> c. curry g This funtion expects its arguments as a pair, but its curried version, curry g, will take them separately - we therefore have to form them into a pair before applying g to them: curry :: ((a,b) -> c) -> (a -> b -> c) curry g x y = g (x,y) curry multiplyUC will be exactly the same function as multiply." OK, I have tried it and it works, but I don't understand the syntax for curry. Until now I have encountered only functions that take the same number of arguments as the function definition or less (partial application), but this syntax looks a bit new to me. curry is supposed to have as its argument one function of type (a,b) -> c and produce another function, but then the second line gives three arguments to curry, the function itself and the variables x and y. What I'm missing here? Thanks a lot, Ángel de Vicente -- http://www.iac.es/galeria/angelv/ High Performance Computing Support PostDoc Instituto de Astrofísica de Canarias --------------------------------------------------------------------------------------------- ADVERTENCIA: Sobre la privacidad y cumplimiento de la Ley de Protección de Datos, acceda a http://www.iac.es/disclaimer.php WARNING: For more information on privacy and fulfilment of the Law concerning the Protection of Data, consult http://www.iac.es/disclaimer.php?lang=en

4 5

A parallelization problem
by wren ng thornton 24 Apr '11

24 Apr '11

Hello all, I have some loopy code I'd like to make parallel, but I'm not sure the best way to go about it. Everywhere else in this program I'm using STM, but that seems pretty heavy-handed for this task, and it looks like the perfect place for `par` and friends. So the 10,000ft view of the code is something like: foldl' step (0,state0) xs step (n,stateN) x = (n+1, k stateN) where k = case lookup n stateN of Nothing -> error "oh noes" Just substateN -> insert (n+1) $ foldl' maybeInsert empty (views substateN x) maybeInsert state (k,v) | p (k,v) = state | otherwise = insert k v state That is, for each x that comes in we add a bunch of things to a map datastructure. The thing that makes the inner loop parallelizable is that we happen to know that none of the new keys will (a) conflict with old keys (as indicated by the fact that we're nesting maps), (b) conflict with the other new keys, nor (c) depend on the other new key/value pairs. Thus, we know that we can reassociate the order of insertions without changing the semantics. So the question is, how do I do it? One issue not apparent in this high-level look at the code is that I'm actually using ST for doing the inner loop efficiently, which is why I'm not just using fromList. So far I'm using an IntMap for storing the state. The STM approach would be to store the IntMap in a TMVar and fire off threads to compute and insert each one of the views. But even as lightweight as Haskell's threads are, I worry that the overhead would swamp the benefits. (Of course, it may be viable to push the ST down far enough to use (fromList . catMaybes). Then we could just construct the list and call par on all the elements. That wouldn't do much to parallelize the insertions themselves, but that shouldn't matter too much I hope. Perhaps I've just answered my own question; any other ideas?) -- Live well, ~wren

1 0