- Glasgow-haskell-users

RE: .hc files for building ghc-6.4.1 on amd64 ?
by Simon Marlow 20 Sep '05

20 Sep '05

On 17 September 2005 22:05, Wilhelm B. Kloke wrote: > I tried it again today. One of my problems was related to gmake > version. > The configure script should warn, if gmake is older than 3.80 (I had > 3.78). I'm using 3.79.1. I do vaguely remember a problem with older versions of gmake - what goes wrong for you? > The building of the .hc bundle fails wirth the following messages: > > ... > tar czf ghc-6.4.1-x86_64-unknown-freebsd-hc.tar.gz `cat > hc-files-to-go` > tar: ghc-6.4.1/ghc/rts/PrimOps.hc: Cannot stat: No such file or > directory > tar: ghc-6.4.1/ghc/rts/StgStartup.hc: Cannot stat: No such file or > directory > tar: ghc-6.4.1/ghc/rts/StgStdThunks.hc: Cannot stat: No such file or > directory > tar: ghc-6.4.1/ghc/rts/Updates.hc: Cannot stat: No such file or > directory > tar: ghc-6.4.1/ghc/rts/Apply.hc: Cannot stat: No such file or > directory > tar: ghc-6.4.1/ghc/rts/Exception.hc: Cannot stat: No such file or > directory > tar: ghc-6.4.1/ghc/rts/HeapStackCheck.hc: Cannot stat: No such file > or directory > tar: ghc-6.4.1/ghc/rts/StgMiscClosures.hc: Cannot stat: No such file > or directory > tar: ghc-6.4.1/libraries/haskell-src/Language/Haskell/Parser.hs: > Cannot stat: No such file or directory gmake: *** [hc-file-bundle] > Fehler 1 Yes, you do actually need to generate these .hc files on the host system. What happened to the build in ghc/rts - could you post the error message? Possibly a 'make -k' in ghc/rts will be enough to get these files. > I tried to go back to ghc/rts to build more files following the > ipattern of the last steps before, but this destroys genapply. Don't quite understand this - could you elaborate? > Further, there are some hsc.c files of length 0 like: > ~/Haskell/fptools-i386amd64/ghc-6.4.1 0$ ls -l > libraries/base/System/*hsc.c > -rw-r--r-- 1 wb ifado 0 17 Sep 21:31 > libraries/base/System/CPUTime_hsc.c > -rw-r--r-- 1 wb ifado 0 17 Sep 21:31 > libraries/base/System/Time_hsc.c This is correct (the C compiler should generate an empty object file when compiling them). It was easier to avoid the special case in the build system when the _hsc.c file is empty. > The building process on the target system fails with messages that > there is no rule to build those files. Also, I needed to have happy > and alex > on the target system, because make boot fails without them (catch 22). When things are working properly, you don't need Happy or Alex on the target system. I suspect things have gone wrong earlier. > A final glitch is that (at least, my) FreeBSD-amd64-6.0 gcc does not > include -I/usr/local/include and -L/usr/local/lib in the search path. > So libgmp is not found. Ok. You can use the libgmp in the tree (this should happen automatically), or you can add the -I/-L flags to the relevant places. This is something we don't handle very well right now, though. Cheers, Simon

2 2

`make' for profiling
by Serge D. Mechveliani 20 Sep '05

20 Sep '05

Dear GHC developers, I have a user's wish for the versions > 6.4.1: the possibility to `make' a project with profiling, without user play with the Cabal version substitution. Regards, ----------------- Serge Mechveliani mechvel(a)botik.ru

1 0

C/Haskell speed comparisons
by Ian Lynagh 20 Sep '05

20 Sep '05

Hi all, Quite often in darcs we want to do a simple loop over a chunk of memory, looking for the first whitespace character or somesuch. Historically this has been done by writing a small snippet of C, but this has a number of downsides (less type safety in the C code, Int/int/HsInt/CInt mismatches, code artificially separated, complicates the Makefiles, have to write C code), so I'd like to replace them with little bits of Haskell. However, it would be easier to convince people that this is the right thing to do if there was no slow-down involved, so I've put some benchmarks together. Hopefully this will help find some cases where ghc is generating poor code, and perhaps help to spot regressions (although it's hard to spot that automatically currently). Essentially each test has n modules, each of which (are supposed to) do the same thing, and it compares certain pairs and it works out how much slower one is than the other on various test sizes. I arbitrarily declared x1.1 to be too slow. The code is at http://urchin.earth.li/~ian/bench/ and the results I get for ghc 6.4.1 are at http://urchin.earth.li/~ian/bench/all.html (interestingly some of the slowdowns I was trying to measure turned out to be speed/ups/). Thanks Ian

1 0

RE: .hc files for building ghc-6.4.1 on amd64 ?
by Simon Marlow 19 Sep '05

19 Sep '05

On 19 September 2005 12:45, Wilhelm B. Kloke wrote: > I tried >> (cd ghc/rts ; gmake all) > after (cd ghc/compiler ; gmake stage=2 ) You should try a 'gmake -k' in ghc/rts, that might be enough to get the required .hc files. > I have further information now. When compiling Linker.c, > the compilation fails, because there is no MAP_32BIT in FreeBSD-amd64. > Perhaps just removing this could make it work. I have no idea > why it is needed on linux-x86_64. MAP_32BIT is quite important: it tells mmap() to return memory in the lower 2Gb of the address space, so that we can assume all symbols have 32 bit addresses. This is how the small memory model on Linux works. I bet FreeBSD has something equivalent (or maybe not?). This is only required for GHCi. You can get a working compiler without needing a working Linker.c. > Probably. But this is in a very early stage. It even happens when > using >> gmake -n boot > (for finding the cause I wanted to know the commands which may have > failed), that two command using $(HAPPY) and $(ALEX) are executed. I > was not > able to locate the point better. Perhaps you could try the first > stages of the crossbuild on suitable system, where these two variable > from > mk/config.mk are defined empty (as was the case on my system after > configure). I would, but I'm kind of busy doing the 6.4.1 release right now. When I have more time, I plan to try the port to FreeBSD/amd64 myself (unless anyone gets there first). Cheers, Simon

1 0

Mac OS X
by Serge D. Mechveliani 19 Sep '05

19 Sep '05

Dear GHC supporters, I have received a letter from a potential DoCon user ************************* > I've downloaded docon 2.08. I am interested in trying to build docon. > I'm using ghc 6.4 intalled on my > Mac OS X machine. > > I've run into an erro and it doesn't look too serious, > but I'm not sure exactly what to do... > > [..] > > Here is the bottom of the file log after running >"make space=-M90m docon >& log": -------------------------------------------- ar -s /Users/ruben/Documents/developer/docon-2/docon/source/export/ libHSdocon.a echo Package {name = "\"docon\"", import_dirs = ["\"/Users/ruben/Documents/developer/docon-2/docon/source/export\""], source_dirs = [], library_dirs = ... /Users/ruben/Documents/developer/docon-2/docon/source/docon.conf -u -g Creating user package database in /Users/ruben/.ghc/powerpc-darwin-6.4/package.conf Reading package info from stdin... ghc-pkg: Line 1: Invalid syntax (no colon after field name) make: *** [docon] Error 1 *********************************** I have replied with the following letter: ---------------------------------------------------- install.txt of docon-2.08 says: -------- It was tested under the Haskell implementation GHC-6.2.2. WARNING: usually, each DoCon version works (literally as it is) only under a single GHC version. On the other hand, it is easy to port to other GHC versions. -------- Mind this warning. In particular, the usage of ghc-pkg (`packages'), changes rapidly with the GHC version. DoCon was tested under Linux. But it is intersesting to see how it will work under Mac OS X. Now, you could try the following alternative ways out. (I) (preferable for me) Install ghc-6.4.1 and ask me to send you a pre-release of docon-2.09. In docon-2.09 + ghc-6.4.1, the package feature is done in a much simpler way (it relies on the Cabal subsystem). With other respects, docon-2.09 is like 2.08. (II) Install ghc-6.2.2, as required by install.txt of docon-2.08, and try to make docon-2.08 under ghc-6.2.2, and write me about the result. -------------------------------------------------- What is your advice, please, concerning the versions of ghc, ghc-pkg, Cabal, as related to Mac OS X. Have we a chance to run (docon-2.09, with docon.cabal) + ghc-6.4.1 under Mac OS X ? Regards, ----------------- Serge Mechveliani mechvel(a)botik.ru

1 0

RE: .hc files for building ghc-6.4 on amd64 ?
by Simon Marlow 17 Sep '05

17 Sep '05

On 12 September 2005 12:53, Wilhelm B. Kloke wrote: > Simon Marlow <simonmar(a)microsoft.com> schrieb: >> On 06 September 2005 15:33, Wilhelm B. Kloke wrote: >> >>> I have a FreeBSD box running FreeSD-amd64-6.0. Is there anybody who >>> can give me a set of .hc files (probably from linux x86_64) to build >>> GHC? I tried to follow the instructions to crossbuild from >>> FreeBSD-i386, but my attempts failed for reasons I don't understand. >> >> I have an account on a box, but haven't yet had time to do the port. >> >> If you let us know where your build got stuck, we might be able to >> help you out. > > Thank you for answering. This is more complicated to do it fast. > I tried to follow the building guide for a new port. IIRC there were > messages related to the difference in range for 64bit constants in the > header files for amd64 and the range which 32bit GHC could handle. > For now I have screwed up both build trees. I am going to restart the > process with fresh copies to give the exact information. > > I was just hoping that a set of Linux-64 .hc files would do > the job for FreeBSD, too. > > BTW, should I hold my breath for 6.4.1? You should certainly bootstrap using a 6.4.1 snapshot source tree, because there have been plenty of amd64-related fixes. There won't be a lot of change between now and the 6.4.1 release. Cheers, Simon

2 1

upgrading glibc
by Gustavo Arroyo Delgado 16 Sep '05

16 Sep '05

Hi, I just want to know, whats the right version of ghc binaries for FC3 at Intel pentium 4 2.4 GHz. I tried it with: ghc-6.4-7.x86_64.rpm ghc-6.4-1.x86_64.rpm but they request the following: [root@mist ~]# rpm -Uvh ghc-6.4-1.x86_64.rpm warning: ghc-6.4-1.x86_64.rpm: V3 DSA signature: NOKEY, key ID aedf7ebb error: Failed dependencies: libc.so.6()(64bit) is needed by ghc-6.4-1.x86_64 libc.so.6(GLIBC_2.2.5)(64bit) is needed by ghc-6.4-1.x86_64 libc.so.6(GLIBC_2.3)(64bit) is needed by ghc-6.4-1.x86_64 libdl.so.2()(64bit) is needed by ghc-6.4-1.x86_64 libgmp.so.3()(64bit) is needed by ghc-6.4-1.x86_64 libm.so.6()(64bit) is needed by ghc-6.4-1.x86_64 libm.so.6(GLIBC_2.2.5)(64bit) is needed by ghc-6.4-1.x86_64 Can I Use #yum install glibc-2.3.5-10.3 with out problems? Thanks in advance. Gustavo Arroyo Delgado ---------------------------------------------------------------- This message was sent using IMP, the Internet Messaging Program.

1 0

Set.lookup?
by Serge D. Mechveliani 16 Sep '05

16 Sep '05

Dear GHC developers, How do you think, maybe, Data.Set also needs to provide `lookup' ? For example, I have data D = D Int String, with the instances Eq and Ord defined by the first coordinate. I do use Map Int String, and for a certain reason, also use Set D. Then, for example, the function check :: D -> Set D -> Bool check (D i name) s = case Set.lookup (D i name) s of Just (D _ nm) -> name == nm _ -> False looks natural, apart that there is no Set.lookup. ----------------- Serge Mechveliani mechvel(a)botik.ru

2 4

RE: FFI: calling Haskell from C++?
by Simon Peyton-Jones 14 Sep '05

14 Sep '05

There's been a bit of traffic about calling Haskell from C++ recently, and I know that calling C++ from Haskell is a frequently asked question. (E.g. Koen Claessen was asking me about it this week.) It also seems to be somewhat platform dependent. (Esp on Windows, I think.) Would someone like to write up a description of how to successfully call out from Haskell to C++ and vice versa, using GHC? If you did, we'd gladly include the result in the user manual, and save others a lot of head scratching. Maybe there are simple things we could do to GHC to make the process simpler. But what it needs is someone who knows C++ well to write a coherent story. Please! Simon ________________________________ From: glasgow-haskell-users-bounces(a)haskell.org [mailto:glasgow-haskell-users-bounces@haskell.org] On Behalf Of Rich Neswold Sent: 13 September 2005 16:45 To: Felix Breuer Cc: glasgow-haskell-users(a)haskell.org Subject: Re: FFI: calling Haskell from C++? On 9/13/05, Felix Breuer <felix(a)fbreuer.de> wrote: $ ghc -fffi Foo.o Foo_stub.o main.cpp main.o(.text+0x22): In function `main': main.cpp: undefined reference to `__stginit_Foo()' main.o(.eh_frame+0x11): undefined reference to `__gxx_personality_v0' collect2: ld returned 1 exit status In main.cpp, you need to define __stginit_Foo(void) like this: extern "C" void __stginit_Foo(void); Specifying the "C" linkage with prevent the compiler from looking for a name-mangled version of the function. To fix the second problem, you need to shut off exception handling (specifying the -fno-exceptions option ought to do this. I don't know how to pass compiler option via ghc, though. -- Rich AIM : rnezzy ICQ : 174908475

3 2

unsafePerformIO and NOINLINE Pragma
by Jan Christiansen 14 Sep '05

14 Sep '05

Hi! I want to analyse the laziness of a data structure. To check how many nodes are constructed I use a global counter. counter :: IORef Int counter = unsafePerformIO (newIORef 0) This counter is increased every time the constructor is called by redefining the constructor OBDD as follows. oBDD low var high = seq (unsafePerformIO (modifyIORef counter (+1))) (OBDD low var high) This works fine. When I compile with optimisations the counter is always set to one no matter how many nodes are constructed. I thought this would be caused by inlining. Therefore I have added two NOINLINE pragmata. {-# NOINLINE counter #-} {-# NOINLINE oBDD #-} Although the counter doesn't work. Is there another optimisation that can cause harm? Is there something wrong with the pragmata? best regards, Jan

2 2