- Haskell-Cafe - Haskell.org

Building shared library to be called from C
by Jack Hill 17 Apr '16

17 Apr '16

Hi, I'm attempting to build a shared library that can be called from C. My end goal is to build a PAM module, but for now I'm working with toy programs so that I can get a feel for building how to build the library. The Haskell part looks like this (in a file called hfact.hs) module Hfact where import Foreign import Foreign.C.Types foreign export ccall fact_hs :: CInt -> CInt fact :: Int -> Int fact x = product [1..x] fact_hs :: CInt -> CInt fact_hs = fromIntegral . fact . fromIntegral I can compile this into a so (I think successfully) with "ghc --make -shared -dynamic -fPIC -o libhfact.so hfact.hs". Now I would like to be able to call this from C. I have a C program (in a file called main.c) as follows: #include <stdio.h> #include "HsFFI.h" #include "hfact_stub.h" int main(int argc, char *argv[]) { int user, answer; scanf("%d", &user); hs_init(&argc, &argv); answer = fact_hs(user); hs_exit(); printf("Result: %d\n", answer); return 0; } Unfortunately, I'm at a loss for how to build and link this against my so. If I try "gcc main.c -lhfact", I get many undefined symbol errors like: /nfshomes/j/jackhill/.stack/programs/x86_64-linux/ghc-7.10.3/lib/ghc-7.10.3/base_HQfYBxpPvuw8OunzQu6JGM/libHSbase-4.8.2.0-HQfYBxpPvuw8OunzQu6JGM-ghc7.10.3.so: undefined reference to `__hscore_get_saved_termios' If I try "ghc maint.c -lhfact" I get: /tmp/ghc19146_0/ghc_4.o: In function `main': ghc_3.c:(.text+0x0): multiple definition of `main' maint.o:maint.c:(.text+0x0): first defined here /tmp/ghc19146_0/ghc_4.o: In function `main': ghc_3.c:(.text+0x61): undefined reference to `ZCMain_main_closure' collect2: error: ld returned 1 exit status How should I be calling Haskell from C? Best, Jack

3 2

is monomorphism restriction necessary?
by Alexey Muranov 16 Apr '16

16 Apr '16

Sorry for the question that probably is already answered somewhere, but i do not get it from the basic documentation i have looked at so far. Is monomorphism restriction really necessary, or is it just a convenience? In the example from A History of Haskell genericLength :: Num a => [b] -> a f xs = (len, len) where len = genericLength xs can't the monomorphism restriction be replaced with just adding the signature f :: Num a => [b] -> (a, a) ? If yes, are there other cases where there is no way to get the desired behavior without enabling the monomorphism restriction? Alexey.

2 4

[ANN] sarsi 0.0.2.0 - A universal quickfix toolkit and his protocol (nvim/vim)
by Alois Cochard 16 Apr '16

16 Apr '16

Hello Café, Here is a new tool which should basically improve the "quick fixing" experience in Haskell with vim/neovim. The aim is not to replace any of the great features provided by integrating the type checker in the editor, but to complement them with a different approach. I explain the motivation there: http://aloiscochard.blogspot.ch/2016/04/quickfix-all-things-with-sarsi.html If you feel like this might be a tool for you, you can jump straight to the README to find the install and usage instructions: https://github.com/aloiscochard/sarsi I did test it extensively with stack, let me know if you have issues using an other tool (I do only pipe stderr which is fine for stack, I suspect this might not be the case for others). Thanks -- *Λ\ois* http://twitter.com/aloiscochard http://github.com/aloiscochard

3 5

Haskell on OpenWRT
by Patrick Pelletier 15 Apr '16

15 Apr '16

Has anyone had any success running Haskell programs on OpenWRT? Specifically, I'm running OpenWRT on x86_64, so processor architecture shouldn't be an issue. However, by default OpenWRT uses musl as its C library, so binaries from a "normal" Linux system wouldn't be compatible with OpenWRT. I attempted to get around this problem by building an OpenWRT image with glibc as the C library. In theory, that ought to solve the problem. In practice, my program (a simple hello-world type program, which runs fine on Ubuntu) hung, printing nothing, using nearly all of one core, and was unkillable by any signal other than SIGKILL. If left alone to run, it eventually used all available memory and then died. I took a look at ldd, to see if there were any clues there. On my Ubuntu 12.04 machine, where I compiled the program (using ghc 7.4.1), I get: ppelletier@patrick64:~/programming/haskell$ ldd contact linux-vdso.so.1 => (0x00007fff36f50000) libgmp.so.10 => /usr/lib/x86_64-linux-gnu/libgmp.so.10 (0x00007f8d49cf8000) libffi.so.6 => /usr/lib/x86_64-linux-gnu/libffi.so.6 (0x00007f8d49af0000) libm.so.6 => /lib/x86_64-linux-gnu/libm.so.6 (0x00007f8d497f3000) librt.so.1 => /lib/x86_64-linux-gnu/librt.so.1 (0x00007f8d495eb000) libdl.so.2 => /lib/x86_64-linux-gnu/libdl.so.2 (0x00007f8d493e7000) libc.so.6 => /lib/x86_64-linux-gnu/libc.so.6 (0x00007f8d49028000) libpthread.so.0 => /lib/x86_64-linux-gnu/libpthread.so.0 (0x00007f8d48e0b000) /lib64/ld-linux-x86-64.so.2 (0x00007f8d49f8d000) On my OpenWRT (trunk) box, I get: root@OpenWrt:~# ldd /tmp/contact linux-vdso.so.1 (0x00007ffd2afcc000) libgmp.so.10 => /usr/lib/libgmp.so.10 (0x00007f27a057a000) libffi.so.6 => /usr/lib/libffi.so.6 (0x00007f27a0373000) libm.so.6 => /lib/libm.so.6 (0x00007f27a0080000) librt.so.1 => /lib/librt.so.1 (0x00007f279fe79000) libdl.so.2 => /lib/libdl.so.2 (0x00007f279fc76000) libc.so.6 => /lib/libc.so.6 (0x00007f279f906000) libgcc_s.so.1 => /lib/libgcc_s.so.1 (0x00007f279f6f4000) libpthread.so.0 => /lib/libpthread.so.0 (0x00007f279f4d9000) /lib64/ld-linux-x86-64.so.2 (0x00007f27a07dc000) This is almost the same, except it has "libgcc_s.so.1", while on the Ubuntu box it does not. I'd welcome any ideas on how to solve (or at least diagnose) this problem. I'd also be interested in any thoughts on the broader question of running Haskell programs on OpenWRT. Thanks, --Patrick

3 8

Re: [Haskell-cafe] Hugs
by Johannes Waldmann 15 Apr '16

15 Apr '16

> the learning process for a student rarely > involves consuming a precise, coherent definition, But it should! I do make a point of embedding references to actual standards in my teaching, mostly in exercises like "what paragraph of the (e.g., Java) language spec. says that the following program is illegal". The intention is: students should a) get used to reading (semi-)formal specifications b) see that concepts from "theory" (lecture, papers) appear verbatim in actual "industrial" specs c) get scared of using software/language that has no specification > .. the nature of functional programming supported by a review of the > theory upon which it is founded: at the very least, the lambda > calculus, preferably along with some basic ideas from type theory. That. And I do start my FP teaching with term rewriting, as a model of first-order programming. (gives you terms as a model of data, and rewrite rule application as a model of pattern matching) - J.W.

1 0

Re: [Haskell-cafe] Hugs
by Doug McIlroy 15 Apr '16

15 Apr '16

> I really do think that pointing students to an unmaintained language > implementation (regardless of the pedagogical reasons) has negative > consequences for the functional programming community as a whole. If it works, maintenance doesn't matter. So, I assume the real concern is that Hugs isn't evolving. To put it in the worst light, this may be read as a complaint that Hugs doesn't keep up with some party line. I would hate to think that the "community as a whole" is that conformist. The disdain for "pedagogical reasons" brushes aside an implicit wakeup call to the community. Hugs is attractive because it is well described and bounded, whereas Haskell realized in GHC lacks a coherent description and presents a myriad of often inscrutable faces. To the extent that the community is defined by such an artifact, it has turned away from educators, not vice versa. Learning how to wrangle a marvelous, but cantankerous, beast should not be confounded with initiation to the insights of functional programming. Doug McIlroy

4 3

Re: [Haskell-cafe] Hugs
by Johannes Waldmann 14 Apr '16

14 Apr '16

> https://twitter.com/headinthebox/status/652834731806052352 This mentions and :: Foldable t => t Bool -> Bool as an example of "abstract nonsense". Well, I will run into teaching this soon myself, though not to freshmen - my students had some Java before but not Haskell. With regards to the type of "and", I will first have them write "data List a = ..." and a concrete "and :: List Bool -> Bool" from scratch. Later, to explain the type of "the true and" I will claim that the corresponding Java method would have type "static Boolean and (Iterable<Boolean> xs)" where Iterable<> is an interface (that the students should know). That's not too bad: xs.iterator() corresponds to toList xs (a lazy stream). "toList" would even be a complete definition for the Foldable instance? (since foldr f z = foldr f z . toList ) So perhaps "Foldable \approx Iterable \approx Enumerable" is a useful guideline. (Yes, there is a difference between a one-argument type class and an interface. But they have some common use cases.) - J.W.

4 3

Re: [Haskell-cafe] how ghc finds libgmp.so when linking
by Johannes Waldmann 14 Apr '16

14 Apr '16

more data: when I compile with -optl -L/usr/local/lib, then the correct library is found, when I put ("ld flags", "-L/usr/local/lib") in ghc-7.10.3/settings, then it's not, when I put ("C compiler link flags", "-L/usr/local/lib") in settings, then it is. So, I can leave it at that. But I still don't know how this came to be. I built this ghc-7.10.3 from source. - J.W.

2 1

Haskell, stack and the Atom editor
by Thomas Koster 14 Apr '16

14 Apr '16

Hi friends, I want to switch to the Atom editor for my Haskell development. I have tried several times over the last few years and given up each time because hardly anything works. I can't stand Emacs so don't bother suggesting it. I am currently using Geany, which has served me well, mostly because it is dead simple so nothing can go wrong. In addition to the usual universal text editor functions, all it does is highlight syntax (albeit with an outdated grammar), runs custom commands to build, displays clickable errors in a separate pane and underlines them in the editor pane. But for larger projects, I want more (better autocompletion, better integration with the build system, better VCS support). But back to Atom, I am having extreme difficulty getting even the basics working. I am using GHC 7.10.2, Cabal 1.22.4.0, Stack 1.0.4. I installed Atom 1.6.1 from their deb package. I installed the "language-haskell" package. This works very well, thankfully, but is the *only* thing that I can get working. I avoided "ide-haskell", even though it looks great, because their readme says it can't build with stack and I rely heavily on stack. Instead, I installed the "build" and "linter" packages and created an ".atom-build" file to build my project with stack. It invokes stack correctly, but I can't get it to recognise any errors: linter always says "No issues". I must have tried twenty different regexes in "errorMatch" but none of them seem to work. I also installed "linter-hdevtools", hoping that that may let me see some errors, but I can't get that to work either. The package is loaded but it appears to do absolutely nothing. strace does show that Atom is trying to exec hdevtools through stack, but I get no output at all within Atom as evidence that something worked or failed. hdevtools appears to have its own problems. I had installed hdevtools (0.1.2.1) with stack: stack install hdevtools I ran it myself using: stack exec --no-ghc-package-path hdevtools -- check Main.hs and it gives me the infamous error: Cabal error: Use of GHC's environment variable GHC_PACKAGE_PATH is incompatible with Cabal. This is strange because "stack exec --no-ghc-package-path env" reports that no such environment variable is defined, as expected. At this point I am ready to give up yet again. Does anybody have any last-minute advice on using Atom to develop with Haskell and Stack? Thanks, Thomas Koster

7 8

[ANN] Vivid 0.2
by amindfv＠gmail.com 14 Apr '16

14 Apr '16

After about a year of blood, sweat, and tears, the new Vivid is out! This gets us to a pretty polished state and updates will now be much more frequent. What is Vivid? It's a library to create music (and other sound) using the SuperCollider synth engine -- a really powerful audio "rendering engine." Vivid lets you create and alter music in real time, or render audio files faster than real time. It's been used in live performances (not only by me!), and I'm very happy with its sound, its power, and its expressivity. Without further ado: http://hackage.haskell.org/package/vivid Happy hacking! Tom

1 0