September 2014 - Haskell-Cafe

Use Haskell shared library with foreign exports with dlopen/dlsym
by Tobias Neumann 10 Sep '14

10 Sep '14

Hello Cafe, I am trying to use a Haskell shared library with foreign exports from Haskell again via dlopen/dlsym. Sadly it segfaults, and the segfaults happen on dlclose during garbage collection points (as figured out by monochrom in #haskell). So right now I can only open a library once and may not dlclose it. Can someone point me to a mistake I made, or is this rather a ghc (7.8.3) bug? Please see attached minimal example. Regards, Tobias test.hs: module Main where import qualified System.Posix.DynamicLinker as DL import Foreign foreign import ccall "dynamic" mkTest :: FunPtr Int -> Int main = do DL.withDL ("./libtest.so") [DL.RTLD_NOW] $ \dl -> do dimPtr <- DL.dlsym dl "test" let test = mkTest dimPtr print test libtest.hs: module Test() where import Foreign foreign export ccall test :: Int test :: Int test = 124 build with: ghc --make -shared -dynamic -fPIC libtest.hs -o libtest.so ghc --make -dynamic test.hs

3 3

Re: [Haskell-cafe] Categorical description for vector-space instance of AdditiveGroup (Maybe a)
by trevor cook 10 Sep '14

10 Sep '14

Here's that vector-space code, formatted better. -- Maybe is handled like the Maybe-of-Sum monoid instance AdditiveGroup a => AdditiveGroup (Maybe a) where zeroV = Nothing Nothing ^+^ b' = b' a' ^+^ Nothing = a' Just a' ^+^ Just b' = Just (a' ^+^ b') negateV = fmap negateV {- Alexey Khudyakov wrote: I looked through vector-space package and found lawless instance. Namely Maybe's AdditiveGroup instance It's group so following relation is expected to hold. Otherwise it's not a group. > x ^+^ negateV x == zeroV Here is counterexample: > let x = Just 2 in x ^+^ negateV x == zeroV False I think it's not possible to sensibly define group instance for Maybe a at all.I see that the problem here is in distinguishing 'Just zeroV' fromNothing. I could fix the Just + Just line to use Nothing instead of JustzeroV when a' ^+^ b' == zeroV, although doing so would require Eq a andhence lose some generality. Even so, the abstraction leak would probablyshow up elsewhere.Hm. -}

1 0

cabal-install fails (osx 10.10)
by Michael Trotter 10 Sep '14

10 Sep '14

I'm having trouble getting cabal-install to work. I used homebrew (brew install cabal-install), which successfully installed the ghc dependency and then attempted to install cabal-install. The problem looks like it's either with my ghc install or maybe an issue with cabal-install and osx 10.10. This is my first haskell experience though, so I'm not really sure how to proceed. Here's the console output: setup: Error: Could not find module: Distribution.Compat.CreatePipe with any suffix: ["dyn_hi"] in the search path: ["dist/build"] cabal: Error: some packages failed to install: Cabal-1.20.0.2 failed during the final install step. The exception was: ExitFailure 1 cabal-install-1.20.0.3 depends on Cabal-1.20.0.2 which failed to install.

3 3

LinkedIn job ad link for openings for several enthusiastic great Haskell developers (geoloc. globally) at Swedish spin-off Functor Group with subsidiaries + three Scala developer (geoloc. Sweden) with spear-edge / very strong qualifications and high-end industry salary for Functor Consulting AB
by Johan Glimming 10 Sep '14

10 Sep '14

Dear Haskell developers, type theory developers, Agda/Epigram/Idris/Coq developers/enthusiasts abound in this community, We are searching the planet for great Haskell and Scala programmers right now for a really interesting spin-off now growing and flourishing after 2.5 years since it was founded. If you are a functional programmer this might be the most exciting job around with e.g. a very exciting customer "as close as a space station as you can get", and industrially entirely unique heavily research-based technology that we have developed now for several years. » https://www.linkedin.com/jobs2/view/19207355 You would help us develop our products further, the new constructive programming paradigm based on type theory carrying languages including domain-specific languages as never done before in the software industry … ( And yes, we are entirely based on (Martin-Löf) type theory, that's what we are all about. And we use Haskell for our development internally almost entirely, with some Ocaml etc on top ... ) We look for three experienced Scala developers (for being based in Sweden, Stockholm-area) at Functor Group AB, Functor AB and Functor Consulting AB. These developers are recruited into Functor Consulting AB which is to become the spear-edge partner to Typesafe in Scandinavia, specifically Sweden, while interacting with our other two teams at Functor AB to a certain extent, focusing on DSL development, DDD and more. We bring type theory to the industry in the new constructive programming paradigm and with a new static analysis technique combined with testing and dependent types, all based on the same core technology … The three Scala developers, for extreme FP consultancy, are offered very competitive Swedish salary, plus added potential for also other benefits via Functor Group AB and the product company Functor AB. Experience from Scala development and documented skills with Scala and related technologies (such as Java) is must be considerable to be considered for these three positions. If you are not based in Sweden, we would assist as much as possible in moving to Stockholm/Uppsala, Sweden, where these three Scala spear-edge developers will be based, with a consultancy manager of considerable experience handling administration and co-ordination in what we expect to be an otherwise quite self-organised business targeting the hardest and most challenging projects at any given time. You can apply on LinkedIn — or, if you prefer, email your resumé, references, credentials, and cover letter to jobs(a)functorgroup.com. » https://www.linkedin.com/jobs2/view/19207355. Best wishes, Johan Glimming — Company website: http://www.functorgroup.com Twitter feeds to follow: @FunctorGroup @Functors @FunctorTech

1 0

Persistent Journal/Event queue Implementation
by Arnaud Bailly 10 Sep '14

10 Sep '14

Hello Cafe, I am looking for a Haskell implementation (or library binding) to some persistent journal/event queue software, something like Apache Kafka. My google-fu is lacking so I was not able to find anything but I might have been doing it wrong. Thanks, -- Arnaud Bailly FoldLabs Associate: http://foldlabs.com

1 0

Re: [Haskell-cafe] FFI question
by Vasili I. Galchin 10 Sep '14

10 Sep '14

On Tue, Sep 9, 2014 at 10:36 PM, Vasili I. Galchin <vigalchin(a)gmail.com> wrote: > Adam, > > Did a find/grep under my posix-realtime directory .. no > definition of MCL_CURRENT, MCL_FUTURE or MCL_CURRENTandFUTURE!! But If > I comment out the case arm MCL_CURRENTandFUTURE in my LockMem.hsc file > and run "cabal build" from the directory where my *.cabal file lives > .. clean compile ... Sigh!??!! > > Vasya > > On Mon, Sep 8, 2014 at 7:36 AM, adam vogt <vogt.adam(a)gmail.com> wrote: >> Hi Vasili >> >> Does the sys/mman.h or another included header actually define >> MCL_CURRENTandFUTURE? >> >> Regards, >> Adam >> >> On Sep 8, 2014 4:45 AM, "Vasili I. Galchin" <vigalchin(a)gmail.com> wrote: >>> >>> Following is a fragment generated by hsc2hs from LockedMem.hsc: >>> >>> #line 93 "LockedMem.hsc" >>> hsc_const (MCL_CURRENTandFUTURE); >>> hsc_fputs (")\n" >>> "", hsc_stdout()); >>> hsc_line (94, "LockedMem.hsc"); >>> hsc_fputs ("\n" >>> "foreign import ccall unsafe \"sys/mman.h mlockall\"\n" >>> " c_mlockall :: CInt -> IO CInt\n" >>> "\n" >>> "\n" >>> "-- | unlock all mapped pages of a process!\n" >>> "unlockAllMemory :: IO () \n" >>> "unlockAllMemory = do\n" >>> " throwErrnoIfMinus1 \"unlockAllMemory\" (c_munlockall)\n" >>> " return ()\n" >>> "\n" >>> "foreign import ccall unsafe \"sys/mman.h munlockall\"\n" >>> " c_munlockall :: IO CInt\n" >>> "", hsc_stdout()); >>> return 0; >>> >>> On Mon, Sep 8, 2014 at 2:11 AM, Vasili I. Galchin <vigalchin(a)gmail.com> >>> wrote: >>> > Hello, >>> > >>> > I need memory refreshing about FFI because I need to fix a bug >>> > (plus I got down a rat hole because of Ukraine war :-(). Following a >>> > snippet of something I wrote: >>> > >>> > -- | lock all of the memory space of a process >>> > lockAllMemory :: LockAllFlags -> IO () >>> > lockAllMemory flags = do >>> > throwErrnoIfMinus1 "lockAllMemory" (c_mlockall cflags) >>> > return () >>> > where >>> > cflags = case flags of >>> > CURRENT -> (#const MCL_CURRENT) >>> > FUTURE -> (#const MCL_FUTURE) >>> > CURRENTandFUTURE -> (#const MCL_CURRENTandFUTURE) >>> > >>> > >>> > it is a file called LockedMem.hsc and the third arm CURRENTandFUTURE I >>> > just added to fix the this bug .. >>> > >>> > MCL_CURRENTandFUTURE ends up in LockedMem_make.c(sp??) but I get a >>> > undeclared nasty message when I try to build LockedMem.hsc... my >>> > bad... . Please help so I can go back to help Ukraine. >>> > >>> > Kind regarsd, >>> > >>> > Vasya >>> _______________________________________________ >>> Haskell-Cafe mailing list >>> Haskell-Cafe(a)haskell.org >>> http://www.haskell.org/mailman/listinfo/haskell-cafe

3 4

2014 haskell cabal update hangs on mac
by Madhu Babu 09 Sep '14

09 Sep '14

I initially installed haskell platform ( 2013 version; 7.6.3 ghc ) on my mac. Everything was working great. Just now saw the haskell platform website again and found new version was released ( Haskell Platform 2014.2.0.0 for Mac OS X, 64bit ). I installed it, and un-installed the older version using uninstall-hs. Now when i type "cabal" or "cabal update" on my terminal, it hangs. Actually when i look into Activity Monitor, i can see that it is invoking some "sh script & possibly some find command" infinitely. I initially guess may be it is building some indexes. but it has been running for an hr or so. I have Xcode 5, gcc and other command line tools installed properly Please advice. I cannot install any other package using cabal.

5 5

Reversing a fixed-length vector
by Cristiano Paris 09 Sep '14

09 Sep '14

Hi, I'm playing around with Type Families so I decided to implement a simple fixed-length Vect of Integers. Here is my straightforward implementation (ghc 7.8.3): https://gist.github.com/anonymous/d838e68ce6a02412859f As you can see, I'm trying to implement a reverse function for my vectors which guarantees that the output vector is the same size as the input one. I tried to express this constraint at the type level. The problem is that I can't have ghc to type check the reverse function in the non-trivial case: _________________ Could not deduce (Plus n1 (S m) ~ S (Plus n1 m)) from the context (n ~ S n1) bound by a pattern with constructor CV :: forall n. Int -> Vect n -> Vect (S n), in an equation for ‘vecReverse’ at vect3.hs:30:13-18 Expected type: Vect (Plus n m) Actual type: Vect (Plus n1 (S m)) _________________ Iit has to do with the fact that the type checker can't deduce that: Plus n1 (S m) ~ S (Plus n1 m) ~ Plus (S n1) m ~Plus n m I tried to insert the following instance to the family: Plus n (S m) = S (Plus n m) but to no avail. Any clue? Thanks. C. -- GPG Key: 4096R/C17E53C6 2010-02-22 Fingerprint = 4575 4FB5 DC8E 7641 D3D8 8EBE DF59 B4E9 C17E 53C6

4 6

Warnings for unhandled exceptions
by Henk-Jan van Tuyl 09 Sep '14

09 Sep '14

L.S., I was reading about the Aardvark language [0] and the robustness they intend to implement in it by, amongst others, catching each possible exception. I wonder, is it possible to implement compiler warnings for unhandled exceptions in Haskell? One of the strong points of Haskell is the type checking and clearly identifying functions with side effects, but exceptions are not reflected by the type of a function. It would be useful if the compiler gave a warning for each type of exception, for each expression, that is not handled by the program. Regards, Henk-Jan van Tuyl [0] http://aardvarklanguage.sourceforge.net/ -- Folding@home What if you could share your unused computer power to help find a cure? In just 5 minutes you can join the world's biggest networked computer and get us closer sooner. Watch the video. http://folding.stanford.edu/ http://Van.Tuyl.eu/ http://members.chello.nl/hjgtuyl/tourdemonad.html Haskell programming --

4 3

Specifying build flags in cabal depencies
by Pedro Vasconcelos 09 Sep '14

09 Sep '14

Hello, I've run across an annoyance with a Haskell project of mine that requires Snap built with SSL: I couldn't find a way to specify the flag "-fopensll" to pass when building the snap-server package. This means that building on a new machine requires either manually installing the right version of snap-server with -fopenssl or re-installing with after the Cabal build. Is there a better way to do this? Thanks, Pedro

2 1