November 2007 - Glasgow-haskell-users

State of parallel GC?
by Ravi Nanavati 16 Nov '07

16 Nov '07

I seem to recall reading in a GHC status report that there was some work going on with parallel garbage-collection. Looking at 6.8.1, it seems pretty clear that it didn't make it. Is there any update about where things are on that front? It isn't an immediately pressing need a Bluespec, but we do get asked every once in a while about parallelizing the Bluespec compiler. One of the technical reasons we're holding off on that front is that it wouldn't win as much as people expect (because garbage collection is usually a decent fraction of a typical run), so we'd be interested if the state-of-play in that area changed. Thanks, - Ravi

6 5

GHC 6.8.1 Build Failure on Mac OS X Intel 10.4.10
by Denis Bueno 15 Nov '07

15 Nov '07

Hi all, Apologies if this has already been reported, or is a symptom of something already reported. I downloaded the GHC source (http://www.haskell.org/ghc/dist/6.8.1/ghc-6.8.1-src.tar.bz2) and attempted to build as follows, bootstrapping with GHC 6.6.1: ./configure ; make -j3 (The error manifests even without `-j3'.) This eventually results in the error [1]: ghc-pkg: dependency readline doesn't exist (use --force to override) But config.log reports: configure:2809: checking for ghc-pkg matching /opt/local/bin/ghc configure:2821: result: /opt/local/bin/ghc-pkg configure:2868: checking whether ghc has readline package configure:2879: result: yes Via MacPorts I have gmp 4.2.2 and readline 5.2.007 installed; I am using XCode 2.5 and gcc 4.0.1. The failure also happens if extralibs is built alongside GHC. I've checked http://hackage.haskell.org/trac/ghc/wiki/X86OSXGhc and googled the error message, but found nothing useful. If there any more useful information I could provide, let me know. -- Denis [1] Here is more of the erroneous output: == Finished recursively making `all' for ways: p ... Registering haskell98-1.0.1.0... Reading package info from "dist/inplace-pkg-config" ... done. Saving old package config file... done. Writing new package config file... done. make stage2 make -C compiler boot stage=2 ../utils/mkdirhier/mkdirhier stage2 gcc -E -undef -traditional -P \ -DIMPORT_DIR='"/Users/Shared/incoming/ghc-6.8.1/libraries/ghc"' \ -DLIB_DIR='"/Users/Shared/incoming/ghc-6.8.1/libraries/ghc"' \ -DINCLUDE_DIR='"/Users/Shared/incoming/ghc-6.8.1/libraries/ghc/include"' \ -DDATA_DIR='"/Users/Shared/incoming/ghc-6.8.1/libraries/ghc"' \ -DHTML_DIR='"/Users/Shared/incoming/ghc-6.8.1/libraries/ghc/html"' \ -DHADDOCK_IFACE='"/Users/Shared/incoming/ghc-6.8.1/libraries/ghc/html/ghc.haddock"' \ -DFPTOOLS_TOP_ABS='"/Users/Shared/incoming/ghc-6.8.1"' \ -x c -DPKG_DEPENDS='hpc bytestring template-haskell readline unix Cabal base haskell98' -DSTAGE='"2"' -I../includes -Iinclude -DPACKAGE=ghc -DVERSION=6.8.1 -DPKG_LIBDIR='"/usr/local/lib/ghc-6.8.1"' -DPKG_DATADIR='"/usr/local/share/ghc-6.8.1"' package.conf.in | \ grep -v '^#pragma GCC' | \ sed -e 's/""//g' -e 's/:[ ]*,/: /g' >package.conf.inplace mkdir stage2 ../utils/ghc-pkg/ghc-pkg-inplace update - --force-files <package.conf.inplace for i in utils basicTypes types hsSyn prelude rename typecheck deSugar coreSyn vectorise specialise simplCore stranal stgSyn simplStg codeGen main profiling parser cprAnalysis ndpFlatten iface cmm nativeGen ghci; do \ ../utils/mkdirhier/mkdirhier stage2/$i; \ done mkdir stage2/utils mkdir stage2/basicTypes mkdir stage2/types mkdir stage2/hsSyn Reading package info from stdin ... done. ghc-pkg: dependency readline doesn't exist (use --force to override) make[2]: *** [../driver/stamp-pkg-conf-ghc] Error 1 make[2]: *** Waiting for unfinished jobs.... mkdir stage2/prelude mkdir stage2/rename mkdir stage2/typecheck mkdir stage2/deSugar mkdir stage2/coreSyn mkdir stage2/vectorise mkdir stage2/specialise mkdir stage2/simplCore mkdir stage2/stranal mkdir stage2/stgSyn mkdir stage2/simplStg mkdir stage2/codeGen mkdir stage2/main mkdir stage2/profiling mkdir stage2/parser mkdir stage2/cprAnalysis mkdir stage2/ndpFlatten mkdir stage2/iface mkdir stage2/cmm mkdir stage2/nativeGen mkdir stage2/ghci make[1]: *** [stage2] Error 2 make: *** [bootstrap2] Error 2

5 6

some links broken in 6.8.1 documentation
by Daniil Elovkov 15 Nov '07

15 Nov '07

Hello folks A quick look at the 6.8.1 user's guide reveals some broken links: 1) Obtaining code coverage, pointing to http://haskell.org/ghc/docs/6.8.1/html/users_guide/hpc.html redirects to http://projects.unsafeperformio.com/hpc.html which tells that it's not found but guesses at another location, where it is found. 2) Concurrent and parallel haskell http://haskell.org/ghc/docs/6.8.1/html/users_guide/lang-parallel.html (and possibly other pages) have links to libraries according to the pre-base-split structure, like .../base/.. instead of .../parallel/..., which results in those links being broken. Thank you -- Daniil Elovkov

3 4

Re: ANNOUNCE: GHC version 6.8.1
by Philip K.F. Hölzenspies 14 Nov '07

14 Nov '07

Dear Ian, all, Even in the current release, I'm still having trouble building GHC on my linux machine. The error is still about DerivedConstants.h. I'm sorry for not following up on my last report, but I'll try and make things right with this one. curl http://www.haskell.org/ghc/dist/6.8.1/ghc-6.8.1-src.tar.bz2 | tar -xj curl http://www.haskell.org/ghc/dist/6.8.1/ghc-6.8.1-src.tar.bz2 | tar -xj cd ghc-6.8.1 sh boot ./configure &> configure.output make &> make.output The outputs are attached to this e-mail. I just wanted to note some of the highlights: The boot script still warns about underquoting (and I still don't know whether this is relevant at all): Booting . Booting libraries/base Booting libraries/directory /usr/share/aclocal/progsreiserfs.m4:13: warning: underquoted definition of AC_CHECK_LIBREISERFS run info '(automake)Extending aclocal' or see http://sources.redhat.com/automake/automake.html#Extending-aclocal Booting libraries/old-time Booting libraries/process Booting libraries/readline Booting libraries/unix The make still fails on the absence of (a rule for) DerivedConstants.h: /usr/local/bin/ghc -H16m -O -istage1/utils -istage1/basicTypes -istage1/types -istage1/hsSyn -istage1/prelude -istage1/rename -istage1/typecheck -istage1/deSugar -istage1/coreSyn -istage1/vectorise -istage1/specialise -istage1/simplCore -istage1/stranal -istage1/stgSyn -istage1/simplStg -istage1/codeGen -istage1/main -istage1/profiling -istage1/parser -istage1/cprAnalysis -istage1/ndpFlatten -istage1/iface -istage1/cmm -istage1/nativeGen -Wall -fno-warn-name-shadowing -fno-warn-orphans -Istage1 -cpp -fglasgow-exts -fno-generics -Rghc-timing -I. -Iparser -package unix -ignore-package lang -recomp -Rghc-timing -H16M '-#include "cutils.h"' -DUSING_COMPAT -i../compat -ignore-package Cabal -c codeGen/CgPrimOp.hs -o stage1/codeGen/CgPrimOp.o -ohi stage1/codeGen/CgPrimOp.hi make[1]: *** No rule to make target `../includes/DerivedConstants.h', needed by `stage1/codeGen/CgProf.o'. Stop. make[1]: Leaving directory `/local/holzensp/src/ghc-6.8.1/compiler' make: *** [stage1] Error 1 Somewhere near the beginning of make's output, where the output for "make boot" is logged, it does seem to have created DerivedConstants.h: ../utils/mkdependC/mkdependC -f .depend -- -O -DTABLES_NEXT_TO_CODE -I. -I../rts -- mkDerivedConstants.c rm -f mkDerivedConstants.o gcc -O -DTABLES_NEXT_TO_CODE -I. -I../rts -c mkDerivedConstants.c -o mkDerivedConstants.o gcc -o mkDerivedConstantsHdr -O -DTABLES_NEXT_TO_CODE -I. -I../rts mkDerivedConstants.o ./mkDerivedConstantsHdr >DerivedConstants.h gcc -o mkGHCConstants.o -O -DTABLES_NEXT_TO_CODE -I. -I../rts -c mkDerivedConstants.c -DGEN_HASKELL gcc -o mkGHCConstants -O -DTABLES_NEXT_TO_CODE -I. -I../rts mkGHCConstants.o ./mkGHCConstants >GHCConstants.h Later in the "make all" phase, there is another mention of DerivedConstants: rm -f mkDerivedConstants.o gcc -O -DTABLES_NEXT_TO_CODE -I. -I../rts -c mkDerivedConstants.c -o mkDerivedConstants.o gcc -o mkDerivedConstantsHdr -O -DTABLES_NEXT_TO_CODE -I. -I../rts mkDerivedConstants.o ./mkDerivedConstantsHdr >DerivedConstants.h gcc -o mkGHCConstants.o -O -DTABLES_NEXT_TO_CODE -I. -I../rts -c mkDerivedConstants.c -DGEN_HASKELL gcc -o mkGHCConstants -O -DTABLES_NEXT_TO_CODE -I. -I../rts mkGHCConstants.o ./mkGHCConstants >GHCConstants.h I don't see the header file being removed anywhere, but it does indeed not show up in the includes directory. When I cheat and, after this failed build, do the following: cd includes make cd .. make the build runs until [42 of 42] and fails, saying: Configuring base-3.0.0.0... Setup: Cannot find ghc at ../../compiler/stage1/ghc-inplace or on the path rm -f base/GNUmakefile cp Makefile.local base if ifBuildable/ifBuildable base; then \ cd base && setup/Setup makefile -f GNUmakefile; \ fi Setup: error reading dist/setup-config; run "setup configure" command? This is where I currently am with trying to build GHC. I hope this serves as a decent bug report. Regards, Philip

6 9

Unpacking across modules
by Scott Dillard 14 Nov '07

14 Nov '07

Hello, What is the interaction between the UNPACK pragma and modules? I've attached a small test case where a datatype is unpacked when it and its associated functions are defined in the same file in which they are used, but it is not unpacked if the definitions are given in another module. There are two versions of the data type, a monomorphic one, defined by: > data Vec3 = Vec3 Double Double Double and a polymorphic, recursive one, defined by > data C a b = C a b With the following typedef to make them equivalent > type Vec3 = C Double (C Double (C Double () )) In the actual file, they're both appropriately annotated with bangs and UNPACK pragmas. The mono datatype is successfully unpacked in both cases, when it is defined in the same file and in a different file. The poly datatype is only unpacked when it is defined in the same file. To see the behavior, compile the attached files with -DMONO_SAME, -DMONO_OTHER, -DPOLY_SAME and -DPOLY_OTHER. In all cases except POLY_OTHER, the program runs in constant space, and in time that is quite competitive with C. (Nice work guys!) However, for POLY_OTHER you will likely have to kill it. The only other flag I used is -O2. Please let me know there's something I'm missing here. Is there a reason why this data type cannot be unboxed across modules? Or is this a bug? Thanks, Scott PS: Also, when the INLINE pragma is used with the Storable instance for the polymorphic data type, it causes heap explosion, even in the same file. Any thoughts here?

2 2

RE: Build failures making GHC 6.8.1 on PPC Mac Os X 10.5
by Corey O'Connor 13 Nov '07

13 Nov '07

(Sorry I couldn't reply to the existing thread; wasn't on the mailing list till recently) I'm also having build problems trying to make GHC 6.8.1 on Mac OS X 10.5 on PPC. Specifically: ../../compiler/stage1/ghc-inplace -package-name parsec-2.1.0.0-hide-all-packages -split-objs -i -idist/build/autogen -idist/build -i. -Idist/build -odir dist/build -hidir dist/build -stubdir dist/build -package base-3.0.0.0 -O -XExistentialQuantification -XPolymorphicComponents -idist/build -H16m -O -O -Rghc-timing -fgenerics -c Text/ParserCombinators/Parsec/Token.hs -o dist/build/Text/ParserCombinators/Parsec/Token.o -ohi dist/build/Text/ParserCombinators/Parsec/Token.hi collect2: ld terminated with signal 10 [Bus error] <<ghc: 335847016 bytes, 43 GCs, 3796398/5983044 avg/max bytes residency (3 samples), 19M in use, 0.00 INIT (0.01 elapsed), 3.07 MUT (23.06 elapsed), 0.54 GC (1.29 elapsed) :ghc>> make[2]: *** [dist/build/Text/ParserCombinators/Parsec/Token.o] Error 1 make[1]: *** [make.library.parsec] Error 2 make: *** [stage1] Error 2 This is on a dual G5 not a PowerBook G4 as previous reports seamed to be from. Anytime I see the linker crash I get nervous. Any clues on where to start debugging? -- -Corey O'Connor

5 5

IArray (IOToDiffArray IOUArray) Bool : Why it doesn't exist ?
by Chaddaï Fouché 13 Nov '07

13 Nov '07

IOUArray Bool exists, and is even a bits array in reality (which makes it really efficient), is there any reason that DiffUArray Bool can't exist ? -- Jedaï

2 1

Binary for Vista 64-bit
by Chris Saunders 13 Nov '07

13 Nov '07

Just wondering if a binary for Vista 64-bit will be released? Regards Chris Saunders

3 2

Re[2]: ghc-6.8.1, +RTS -N2
by Bulat Ziganshin 13 Nov '07

13 Nov '07

Hello Johannes, Tuesday, November 13, 2007, 12:08:20 PM, you wrote: >> i suggest you to test it on one-threaded CPU too > I just did. The behaviour is identical > (works with -N1, segfaults with -N2) > Any ideas? - Best regards, J.W. no. just publish this into maillist, it's important to identify the problem well, in my practice the only way to make GHC crash (except for errors in GHC itself) is to use unsafePerformIO, in particular for creation global variables -- Best regards, Bulat mailto:Bulat.Ziganshin@gmail.com

2 1

ghc-6.8.1, +RTS -N2
by Johannes Waldmann 13 Nov '07

13 Nov '07

what could be the reason for my program (compiled with -threaded) crashing with +RTS -N2 it seems to work fine with "-N1". I compiled with ghc-6.8.1 (installed from binary package). I am using forkIO, some MVars, some IORefs, some Channels. It seems -prof -threaded does not work: ghc-6.8.1: combination not supported: Threaded/Profiling When I run the executable under gdb I get Program received signal SIGSEGV, Segmentation fault. [Switching to Thread 0xb6afeb90 (LWP 7801)] 0x08138d8c in checkBlackHoles () This behaviour occurs both on a dual core PC and on an single core hyperthreaded PC, both running openSuse-10.3 I cannot tell whether this is an regression from 6.6 (I cannot easily backport my code). Any hints appreciated, Johannes Waldmann.

1 0