July 2006 - Glasgow-haskell-users

(Repost) Replacement for GMP as Bignum: ARPREC? Haskell?; OS X and OpenSSL
by p.tanski＠gmail.com 30 Jul '06

30 Jul '06

(Sorry for the repost: I needed to correct a few typos--thanks Seth--and clarify a few things.) GHC Task Ticket # 601 suggests replacing GMP with OpenSSL's Bignum library, BN. I have two questions concerning this: (1) Why not use the ARbitrary PRECision Computation Package (ARPREC) by David Bailey, Yozo Hida, Karthik Jeyabalan, Xiaoye Li and Brandon Thompson? Here is a reference web page: http://crd.lbl.gov/~dhbailey/mpdist/ ARPREC is written in C++ but supports calls from C (see include/arprec/c_mp.h in the distribution directory). ARPREC is very fast and supports more complex mathematics than BN. The Licensing for ARPREC is not a problem: essentially similar to BSD3. Note that there would be a basic configuration fix for building OpenSSL's BN library on OS 10.4: you would probably--I had to, myself--have to temporarily move the default installation of /usr/lib/libssl0.9.dylib (and variants) if you want to create a static (.a) library because Apple's ln links dynamic libraries in preference to static libraries. (2) A much more aesthetic solution would be to replace any external Bignum library with a pure Haskell library. There are many problems with building a truly efficient and fast Bignum library in Haskell but challenges like that are how languages evolve... Both GHC and nhc98 seem to use GMP, OCaml uses a custom bignum library written by Xavier Leroy and even MLton uses GMP. Robert Dockins wrote a type-based library for arbitrary that can (slowly) handle big numbers (up to 10^15), see http://www.eecs.tufts.edu/~rdocki01/typenats.html I don't know what the constraints on making such native Integer support work in GHC but my comment on the evolution of Haskell was not meant badly: native Integer support would require modifications to how Integers are handled in memory. (I would be glad to lend whatever support I could on this, but it seems like a big project.) -Peter Tanski

3 2

Re: Re: Rebuilding GHC on Mac OSX PPC
by p.tanski＠gmail.com 30 Jul '06

30 Jul '06

Sorry to flood your mailboxes with my junk, but I had a few questions on the configuration and Makefiles for building GHC on OS 10.4: The odd memory bugs (wierd malloc errors) seem to remind me of errors using pthreads... in Joel Reymont's email, I noticed that ghc is given the -threaded option: ../../ghc/compiler/stage1/ghc-inplace -o stage2/ghc-6.4.3 ... ... -threaded -package readline -DUSE_READLINE - I have not had any luck using -threaded with GHC on my own programs, but I do not know what options GHC passes to gcc with -threaded. If GHC -threaded means it passes gcc the -pthread option when compiling -fvia-C, the problem might be that Apple's gcc does not recognize the -pthread flag (you have to link -lpthread manually, preferably after the object files). If I can get the configuration correct--or whatever parameters you think you might want--I will get a clean distribution and try to build ghc-6.4.3 tomorrow (I am using a 1.25 MHz PPC). -Peter Tanski

1 0

GHC 6.4.3 is stalled
by Simon Peyton-Jones 29 Jul '06

29 Jul '06

Folks GHC 6.4.2's threaded runtime system does not work right on Solaris MacOSX Possibly FreeBSD We'd love to fix these bugs and release 6.4.3, but we are stalled because we don't have easy access to these platforms, nor do we have detailed expertise in their intimate details (which is where the bugs will be lurking). If you care about these platforms, would you like to lend a hand? We probably need more help than simply "here's access to the platform", but we'd give you very strong support if you were willing to look into it. Failing that, I think we'll have to stick with 6.4.2. (And the same bugs may well show up in GHC 6.6, which we hope to release before ICFP.) Simon

8 12

Chasing a memory leak
by Rich Fought 28 Jul '06

28 Jul '06

I'm getting into the weeds of a GHC-compiled program, and need a little help. I'm trying to chase down a memory leak and Valgrind gives me the following: ==24390== 54,844 bytes in 639 blocks are definitely lost in loss record 69 of 69 ==24390== at 0x400446D: malloc (vg_replace_malloc.c:149) ==24390== by 0x826337E: (within myprog) From the link map of my program this address is in the function GHCziBase_zdszddmmin_info I can't seem to find any similar reference in the GHC source or in any of my assembly files, so I'm assuming it's something the compiler puts together internally for the GHC runtime. What does this function do, or how can I find out what it does? Thanks, Rich

2 2

GHC compile times (was Re: GHC 6.4.3 is stalled)
by Joel Reymont 27 Jul '06

27 Jul '06

On Jul 25, 2006, at 1:34 PM, Christian Maeder wrote: > On our solaris sparc machine compiling our main binary (optimized) > takes > 3h:38min whereas (only) 55min under linux. At least our sparcs may die > out sooner or later. Interestingly enough, it takes the 3+ hours to compile GHC 6.5 on my Intel Core-Duo Mac. Why would there be such a huge difference between Linux and Mac OSX on the same architecture? How can I pinpoint the bottleneck? Thanks, Joel -- http://wagerlabs.com/

8 12

Building NewBinary with ghc 6.5
by Joel Reymont 26 Jul '06

26 Jul '06

Is there something that looks particularly wrong below? I'm on Mac OSX Intel. I'm trying to build crypto which depends on NewBinary which depends on lang (removed). --- ghc-pkg -l /usr/local/lib/ghc-6.5/package.conf: ALUT-2.0, Cabal-1.1.4, GLUT-2.0, HUnit-1.1, OpenGL-2.1, QuickCheck-1.0, base-1.0, fgl-5.2, (ghc-6.5), haskell-src-1.0, haskell98-1.0, mtl-1.0, network-1.0, parsec-2.0, readline-1.0, rts-1.0, stm-1.0, template-haskell-1.0, time-1.0, unix-1.0 junior:~/work/Haskell/NewBinary joelr$ sudo runhaskell Setup.lhs install Installing: /usr/local/lib/NewBinary-0.1/ghc-6.5 & /usr/local/bin NewBinary-0.1... Registering NewBinary-0.1... Reading package info from ".installed-pkg-config" ... done. ghc-pkg: invalid package identifier: -- http://wagerlabs.com/

2 4

Transformational Patterns?
by Pete Kazmier 25 Jul '06

25 Jul '06

After reading the paper "Pattern Guards and Transformational Patterns" by Martin Erwig and Simon Peyton Jones, I'm left wondering about the status of transformational patterns? Can we expect to see these at some point in GHC? Or have they gone by the wayside in favor of some other alternative? When I had finished reading the paper, I was disappointed to find out that they were not implemented as I was convinced of the merits based on the arguments presented in the paper. Thanks, Pete [1] http://research.microsoft.com/~simonpj/Papers/pat.htm

2 1

-fasm vs. -fvia-c
by Joel Reymont 25 Jul '06

25 Jul '06

On Jul 25, 2006, at 2:09 PM, Duncan Coutts wrote: > You can also cut compile times by compiling fewer non-core > libraries and > by using -fasm rather than -fvia-C. I heard recently that GHC does not generate code that gcc likes and so there's a push to improve asm generation. What stage is that push at? Do I cripple my GHC build in any way by using asm as opposed to via-c? Thanks, Joel -- http://wagerlabs.com/

2 1

Internships on GHC and Haskell at MSR Cambridge
by Simon Peyton-Jones 25 Jul '06

25 Jul '06

Gentle Haskellers Would you be interested in working for three months at Microsoft Research, Cambridge, on a project related to GHC or Haskell? MSR Cambridge now takes interns *year-round*, not just in the summer months. Simon Marlow and I are keen to attract motivated and well-qualified folk to work with us on our research, and on improving or developing GHC. If you or one of your students is interested, you can find more details here http://hackage.haskell.org/trac/ghc/wiki/Internships Our next empty slot is the Oct-Dec period, but you are welcome to apply for some later period. Simon PJ PS: There are lots of other programming-language folk at MSR (F#, security, probabilistic languages...), and stuff beyond that (machine learning, systems and networks...); see http://research.microsoft.com/aboutmsr/labs/cambridge/

1 0

Text.Regex.Posix regcomp
by Rich Fought 24 Jul '06

24 Jul '06

Hello, I have a server application that I am building using GHC 6.4 (yes, an update to 6.4.2 is on the horizon, but not in the immediate future - unless it fixes this problem :) ) under MSYS/MinGW. Things work pretty well, but under stress testing the app eventually throws a Windows memory access violation (c0000005) exception out of the exact same location in the regcomp function every time. When it occurs is not deterministic - it could be after 20 requests or 2000 (usually higher, though). Various Windows debuggers shows it as a memory read error. regcomp is called several times with each incoming request. I'm pretty sure it's not a multi-threading issue, as there is only one client sending requests over and over again. Can anyone suggest some ideas or debugging tips? TIA, Rich

4 8