- Haskell-Cafe - Haskell.org

MonadError vs Control.Exception
by Станислав Черничкин 06 May '12

06 May '12

Hi, guys, I'm interested in best practices in using of each approach. Personally I like MonadError because it is more explicit and Control.Exception-s becomes really ugly in complex scenarios. Here an example to illustrate my idea: http://hackage.haskell.org/packages/archive/network-conduit/0.4.0/doc/html/…. All network-conduit functions can raise IOException because of Socket-related stuff (and actually entire network-conduit is written to deal with IOException correctry) + Bind-port can raise ErrorCall (came from error "bindPort: addrs is empty"). User has to deal with both, but have no evidence on none of them from type signatures or documentation. Moreover, ErrorCall in this case can be caused by IOException (we try to bind only available address, we get IOException exception, we have no more address to try and raise ErrorCall), and I really don’t understand, why it should be treated in a different way. So, I hate exceptions, I blame it, I think exceptions is junk came from OO-world and horrible languages like C# or even more horrible like Java, and it should be wiped out from Haskell with fire. But it is only my humble opinion. Reality shows that Haskell provides extensive support for exceptions, it included in base libraries and everyone uses it, but MonadError is modest part of mtl, and don’t even have “bracket” and other useful functions. That is why I’m really interested in any your ideas and experience on using these approaches. Cheers, Stasik.

3 3

Learn you
by Kazu Yamamoto 06 May '12

06 May '12

Hello cafe, Translating "Learn You a Haskell for Great Good" into Japanese was finished and will be published on 22 May. I guess it's worth watching its cover page: http://www.amazon.co.jp/%E3%81%99%E3%81%94%E3%81%84Haskell%E3%81%9F%E3%81%A… There are two translators: Tanaka is the author of "peggy" and Muranushi is the author of "Monadius". Regards, --Kazu

14 17

Issues with installing packages
by Graham Berks 06 May '12

06 May '12

New to haskell and having issues resolving packages, have the following output below and can't seem to resolve it. Text seems to have diff hash dependancies when I try and instal a diff version of 0.11.2.0 etc Any ideas on what todo ?? Thanks $ ghc-pkg check There are problems in package scion-browser-0.2.8: dependency "text-0.11.2.0-2b6cea8526e93af24bb0154400fd64a6" doesn't exist There are problems in package persistent-template-0.9.0.1: dependency "text-0.11.2.0-2b6cea8526e93af24bb0154400fd64a6" doesn't exist There are problems in package persistent-sqlite-0.9.0.1: dependency "text-0.11.2.0-2b6cea8526e93af24bb0154400fd64a6" doesn't exist There are problems in package persistent-0.9.0.3: dependency "text-0.11.2.0-2b6cea8526e93af24bb0154400fd64a6" doesn't exist There are problems in package aeson-0.6.0.2: dependency "text-0.11.2.0-2b6cea8526e93af24bb0154400fd64a6" doesn't exist There are problems in package path-pieces-0.1.1: dependency "text-0.11.2.0-2b6cea8526e93af24bb0154400fd64a6" doesn't exist There are problems in package hashable-1.1.2.3: dependency "text-0.11.2.0-2b6cea8526e93af24bb0154400fd64a6" doesn't exist There are problems in package conduit-0.4.1.1: dependency "text-0.11.2.0-2b6cea8526e93af24bb0154400fd64a6" doesn't exist There are problems in package blaze-html-0.4.3.4: dependency "text-0.11.2.0-2b6cea8526e93af24bb0154400fd64a6" doesn't exist There are problems in package blaze-builder-0.3.1.0: dependency "text-0.11.2.0-2b6cea8526e93af24bb0154400fd64a6" doesn't exist There are problems in package attoparsec-0.10.1.1: dependency "text-0.11.2.0-2b6cea8526e93af24bb0154400fd64a6" doesn't exist There are problems in package buildwrapper-0.5.2: dependency "aeson-0.6.0.2-f5264d3642c44cd43d87b542486e49c9" doesn't exist dependency "attoparsec-0.10.1.1-d3ad0453e3127b6c51ccb14a5cb6f38d" doesn't exist dependency "unordered-containers-0.2.1.0-9b5d4d7fe2d79561a684ee5184ea5fd0" doesn't exist There are problems in package hoogle-4.2.11: dependency "blaze-builder-0.3.1.0-adf48413bb3bfeef99ef5e18de097dbd" doesn't exist dependency "conduit-0.4.1.1-6c1a48d4c94e05132c53b85e7ca48cd9" doesn't exist dependency "parsec-3.1.1-d3c0b4413115a5e82a1a5cf9b0b35fee" doesn't exist There are problems in package warp-1.2.0.1: dependency "blaze-builder-0.3.1.0-adf48413bb3bfeef99ef5e18de097dbd" doesn't exist dependency "conduit-0.4.1.1-6c1a48d4c94e05132c53b85e7ca48cd9" doesn't exist There are problems in package wai-1.2.0.1: dependency "blaze-builder-0.3.1.0-adf48413bb3bfeef99ef5e18de097dbd" doesn't exist dependency "conduit-0.4.1.1-6c1a48d4c94e05132c53b85e7ca48cd9" doesn't exist There are problems in package network-conduit-0.4.0: dependency "conduit-0.4.1.1-6c1a48d4c94e05132c53b85e7ca48cd9" doesn't exist There are problems in package blaze-builder-conduit-0.4.0.1: dependency "blaze-builder-0.3.1.0-adf48413bb3bfeef99ef5e18de097dbd" doesn't exist dependency "conduit-0.4.1.1-6c1a48d4c94e05132c53b85e7ca48cd9" doesn't exist There are problems in package network-2.3.0.5: dependency "parsec-3.1.1-d3c0b4413115a5e82a1a5cf9b0b35fee" doesn't exist There are problems in package http-types-0.6.10: dependency "blaze-builder-0.3.1.0-adf48413bb3bfeef99ef5e18de097dbd" doesn't exist There are problems in package case-insensitive-0.4.0.1: dependency "hashable-1.1.2.3-037c817e51d5d10d89306e9472e2e0bd" doesn't exist There are problems in package test-framework-th-0.2.2: dependency "test-framework-0.6-7d2482720de09487769f301ebc001970" doesn't exist There are problems in package test-framework-quickcheck2-0.2.12.1: dependency "test-framework-0.6-7d2482720de09487769f301ebc001970" doesn't exist There are problems in package test-framework-quickcheck-0.2.8: dependency "test-framework-0.6-7d2482720de09487769f301ebc001970" doesn't exist There are problems in package test-framework-hunit-0.2.7: dependency "test-framework-0.6-7d2482720de09487769f301ebc001970" doesn't exist The following packages are broken, either because they have a problem listed above, or because they depend on a broken package. scion-browser-0.2.8 persistent-template-0.9.0.1 persistent-sqlite-0.9.0.1 persistent-0.9.0.3 aeson-0.6.0.2 path-pieces-0.1.1 hashable-1.1.2.3 conduit-0.4.1.1 blaze-html-0.4.3.4 blaze-builder-0.3.1.0 attoparsec-0.10.1.1 buildwrapper-0.5.2 hoogle-4.2.11 warp-1.2.0.1 wai-1.2.0.1 network-conduit-0.4.0 blaze-builder-conduit-0.4.0.1 network-2.3.0.5 http-types-0.6.10 case-insensitive-0.4.0.1 test-framework-th-0.2.2 test-framework-quickcheck2-0.2.12.1 test-framework-quickcheck-0.2.8 test-framework-hunit-0.2.7 unordered-containers-0.2.1.0 pool-conduit-0.1.0.1 resource-pool-0.2.1.0 simple-sendfile-0.2.3 cgi-3001.1.7.4 haskell-platform-2011.4.0.0 HTTP-4000.1.2

2 1

unsafePerformIO usage
by Serge D. Mechveliani 06 May '12

06 May '12

People, can anybody explain what incorrect may happen with the following program which uses exchange through two Unix pipes and unsafePerformIO ? The pipes are created by the Unix commands mkfifo toA; mkfifo fromA gcc iface.c compiles the C program to a.out. ghc --make Main compiles the Haskell program by GHC to ./Main Run on the second terminal > ./iface Run on the first terminal > ./Main It prints the result to the screen. In Main.hs, the function iface :: String -> String is defined via hPutStr, hGetLine, hFlush, unsafePerformIO. It outputs str :: String to the pipe toA; the process iface (programmed in C) waits for delay and returns the respond string through the pipe fromA; iface inputs this respond string. So far, I put the respond string to be the same string, I think, this is not essential, but another map can be considered. Assume that the parallel computation is not allowed. In the below `main', I tried f ss where ss = [iface $ shows i "\n" | i <- [1 .. 4]], f [s1, s2, s3, s4] = [s3++s1, s3++s2, s1++s2], and it works correct. The question is: what a simple definition is possible for f which will evaluate incorrect? Is this true that unsafePerformIO is _safe_ for all f in this context? The `delay' constant can be edited in fifo.c. See both sources below. Please, copy the respond to mechvel(a)botik.ru Thanks, ------ Sergei mechvel(a)botik.ru ------------------------------------------------------------------------ import System.IO (IOMode(..), IO(..), Handle, openFile, hPutStr, hGetLine, hFlush) import System.IO.Unsafe (unsafePerformIO) dir = showString "/home/mechvel/haxiom/unsafeTest/" -- openFile :: FilePath -> IOMode -> IO Handle -- hPutStr :: Handle -> String -> IO () -- hGetLine :: Handle -> IO String toA_IO = openFile (dir "toA") WriteMode :: IO Handle fromA_IO = openFile (dir "fromA") ReadMode toA = unsafePerformIO toA_IO fromA = unsafePerformIO fromA_IO initIO :: IO (Handle, Handle) initIO = do to <- toA_IO from <- fromA_IO return (to, from) ifaceIO :: (Handle, Handle) -> String -> IO String ifaceIO (h1, h2) str = do hPutStr h1 str hFlush h1 str' <- hGetLine h2 return str' iface :: String -> String iface str = unsafePerformIO $ ifaceIO (toA, fromA) str main = putStr $ shows [s3++s1, s3++s2, s1++s2] "\n" where ss@[s1, s2, s3, s4] = [iface $ shows i "\n" | i <- [1 .. 4]] -- The C program fifo.c --------------------------------------------- #include <stdio.h> #include <string.h> #define BOUND 64 static char str[BOUND]; main() { int l, i; long j; FILE *toA, *fromA; toA = fopen("/home/mechvel/haxiom/unsafeTest/toA", "r"); if (toA == NULL) {perror("fopen(toA, r) failed: "); return;}; fromA = fopen("/home/mechvel/haxiom/unsafeTest/fromA", "w"); if (fromA == NULL) {perror("fopen(fromA, w) failed: "); return;}; for (;;) { if (fgets(str, BOUND, toA) == NULL) { perror("fgets(str, bound, toA) failed: "); return; }; j = 20000000; // delay. edit it while (j > 0) {j = j-1;}; fputs(str, fromA); fflush(fromA); }; return; }

1 0

library conflicts and how to resolve them
by Graham Berks 06 May '12

06 May '12

Attempting $ cabal install persistent Resolving dependencies... In order, the following would be installed: aeson-0.6.0.2 (reinstall) changes: mtl-2.1.1 -> 2.0.1.0 persistent-0.9.0.3 (new package) cabal: The following packages are likely to be broken by the reinstalls: buildwrapper-0.5.2 Use --force-reinstalls if you want to install anyway. Any ideas what is going on ? I believe aeson-0.6.0.2 is already installed. And as far as I can tell the build of aeson-0.6.0.2 doesn't require that specific version of mtl. Am I missing something ? Is there a way to force ghc to ignore this ?? Thanks

3 2

Haskell and arrays
by Morten Olsen lysgaard 05 May '12

05 May '12

I'm a bit confused on what the standard is when it comes to arrays and performance in Haskell. Say I have a function that takes an n dimensional array. The array only contains some primitive type. The function mutates the array in an deterministic but chaotic pattern. The new value of an element might depend on other values in the array, and the order of update might depend on what values are found. How would I implement this in an efficient fashion in Haskell. Is the array package the way to go? Would vector with some custom indexing scheme work well? The array package provides n dimensional indexing, vector doesn't. Does vector perform better than array? Is it realistic to approach the performance of a naive c implementation? What steps would need to be taken in order to achieve that? I know there are many question here, but I don't expect perfect answers to them. I just need to bootstrap my map of numerics in Haskell. PS: An example of the above function could be Gaussian elimination of a matrix. -- Morten Olsen Lysgaard NTNU-IT - Orakeltjenesten NTNU - Physics and Mathematics

3 2

haskellonline.org, a web frontend for ghc
by Gintautas Miliauskas 05 May '12

05 May '12

Hello, check out http://haskellonline.org, an online Haskell typechecker. It is essentially a thin web wrapper over ghc, but it takes some friction out of learning Haskell, if only because no local setup is required. It is a weekend's work, so do not be too harsh, but feedback is welcome! I have also written up an accompanying blog post [1]. [1] http://blog.miliauskas.lt/2012/05/online-haskell-typechecker.html Best regards, -- Gintautas Miliauskas

2 1

Generalizing (++) for monoids instead of using (<>)
by aditya bhargava 04 May '12

04 May '12

After asking this question: http://stackoverflow.com/questions/9963050/standard-way-of-joining-two-data… I found out that the new infix operator for `mappend` is (<>). I'm wondering why ghc 7.4 didn't generalize (++) to work on monoids instead. To me, (++) is much more clear. (<>) means "not equal to" for me. Can anyone shed light on this decision? Adit -- adit.io

9 9

ANN: meta-par, meta-par-accelerate and friends -- heterogeneous parallel programming
by Ryan Newton 04 May '12

04 May '12

Hi all, I'm happy to announce a release of meta-par, a parallel programming library that is a successor to monad-par. - http://hackage.haskell.org/package/abstract-par - http://hackage.haskell.org/package/monad-par - http://hackage.haskell.org/package/meta-par - http://hackage.haskell.org/package/abstract-par-accelerate - http://hackage.haskell.org/package/meta-par-accelerate The distributed programming components aren't released yet, but this blog posts describes how to use the above packages to do hybrid CPU/GPU programming: http://parfunk.blogspot.com/2012/05/how-to-write-hybrid-cpugpu-programs.html Cheers, -Ryan

1 0

conduit: Finalize field in PipeM
by Paolo Capriotti 04 May '12

04 May '12

Hi, I'm trying to write a function to convert a conduit to a Pipe (from pipes-core), and I'm having trouble understanding why the `Finalize` field in a `PipeM` constructor returns `r`. This makes it impossible to create it from the corresponding `M` constructor in pipes-core, since `M` includes an exception handler which is not guaranteed to run fully, hence may not provide a return value. `HaveOutput` presents a similar problem. >From a cursory look at conduit's code, it doesn't look like the return value of `Finalize` is ever used, and it seems that the conduits that actually manage to define it either have `()` as return type or just throw an exception. The definition of `lift` for the `MonadTrans` instance duplicates the base monad action there, which doesn't look quite right to me (isn't that supposed to contain a _cleanup_ action?). Shouldn't the constructor be changed to something like `PipeM (m (Pipe i o m r)) (Finalize m ())` or am I completely off base here? Thanks. BR, Paolo

2 2