From Malcolm.Wallace at cs.york.ac.uk Thu Mar 1 05:13:03 2007 From: Malcolm.Wallace at cs.york.ac.uk (Malcolm Wallace) Date: Thu Mar 1 05:06:27 2007 Subject: [cvs-nhc98] patch applied (nhc98): some doc updates, including note about working on 64-bit machines Message-ID: <20070301101303.GA20166@cvs.haskell.org> Thu Mar 1 02:09:02 PST 2007 Malcolm.Wallace@cs.york.ac.uk * some doc updates, including note about working on 64-bit machines M ./docs/download.html -1 +3 M ./docs/extensions.html -1 +2 M ./docs/history98.html +1 M ./docs/index.html -2 +7 M ./docs/install.html -2 +4 M ./docs/libraries.html +3 M ./docs/maillist.html -2 +2 M ./docs/status.html -1 +1 M ./docs/york-release.html -4 +22 From sven.panne at aedion.de Thu Mar 1 06:00:43 2007 From: sven.panne at aedion.de (Sven Panne) Date: Thu Mar 1 05:54:05 2007 Subject: [cvs-nhc98] Status of nhc98 on 64bit platforms In-Reply-To: <20070227154524.06c7a10a.Malcolm.Wallace@cs.york.ac.uk> References: <200702211518.46450.sven.panne@aedion.de> <20070222101229.46350d44.Malcolm.Wallace@cs.york.ac.uk> <20070227154524.06c7a10a.Malcolm.Wallace@cs.york.ac.uk> Message-ID: <200703011200.43332.sven.panne@aedion.de> On Tuesday 27 February 2007 16:45, Malcolm Wallace wrote: > I have pushed a patch today that aims to do exactly as I suggested. Can > you attempt to configure-and-build with this in place, and report any > failures? Things look a bit better now (only a few warnings), but still not OK: /home/panne/local/nhc98/script/nhc98 -c +CTS -part -redefine -CTS -P../LowB/ -P../Ratio -P../PreludeList -P../PreludeText -P../PreludeIO -prelude -o /home/panne/local/nhc98/targets/x86_64-Linux/obj/prelude/Prelude/Enum_Double.o Enum_Double.hs /home/panne/local/nhc98/script/nhc98: line 417: 26896 Segmentation fault $COMP $RUNFLAGS $COMPFLAGS $HINCDIRS -I. $UNLIT -P${NHC98INCDIR} $HSFILE $POSTGRFILE $TMPHIFILE $TMPCPPASFILE I'll take a look at this in the next few days... Cheers, S. From Sven.Panne at aedion.de Thu Mar 1 07:11:37 2007 From: Sven.Panne at aedion.de (Sven Panne) Date: Thu Mar 1 07:05:01 2007 Subject: [cvs-nhc98] patch applied (nhc98): Fixed ARCH confusion Message-ID: <20070301121137.GA23922@cvs.haskell.org> Thu Mar 1 04:03:30 PST 2007 sven.panne@aedion.de * Fixed ARCH confusion When ghc_floats.c is compiled, we do *not* use ARCH flags like -m32, because this code is linked to code compiled by GHC, which uses the native ABI. The correct way to compile ghc_floats.c would be using GHC itself, because then ABI compability is guaranteed. Due to my limited knowledge of nhc98's build system, I leave this to Malcolm... M ./src/compiler98/Makefile -1 +1 From Sven.Panne at aedion.de Thu Mar 1 08:40:21 2007 From: Sven.Panne at aedion.de (Sven Panne) Date: Thu Mar 1 08:33:43 2007 Subject: [cvs-nhc98] patch applied (nhc98): Use ARCH when compiling C files in the Prelude Message-ID: <20070301134021.GA26671@cvs.haskell.org> Thu Mar 1 05:38:58 PST 2007 sven.panne@aedion.de * Use ARCH when compiling C files in the Prelude M ./src/prelude/Makefile.common -1 +1 From Malcolm.Wallace at cs.york.ac.uk Thu Mar 1 08:58:40 2007 From: Malcolm.Wallace at cs.york.ac.uk (Malcolm Wallace) Date: Thu Mar 1 08:55:39 2007 Subject: [cvs-nhc98] patch applied (nhc98): Fixed ARCH confusion In-Reply-To: <20070301121137.GA23922@cvs.haskell.org> References: <20070301121137.GA23922@cvs.haskell.org> Message-ID: <20070301135840.24e8940b.Malcolm.Wallace@cs.york.ac.uk> > Thu Mar 1 04:03:30 PST 2007 sven.panne@aedion.de > * Fixed ARCH confusion > > When ghc_floats.c is compiled, we do *not* use ARCH flags like -m32, > because this code is linked to code compiled by GHC, which uses the > native ABI. *Nod*. > The correct way to compile ghc_floats.c would be using GHC itself, > because then ABI compability is guaranteed. Due to my limited > knowledge of nhc98's build system, I leave this to Malcolm... That would make sense. Regards, Malcolm From Malcolm.Wallace at cs.york.ac.uk Thu Mar 1 09:14:01 2007 From: Malcolm.Wallace at cs.york.ac.uk (Malcolm Wallace) Date: Thu Mar 1 09:07:22 2007 Subject: [cvs-nhc98] patch applied (nhc98): Use ghc rather than gcc to compile ghc_floats.c Message-ID: <20070301141401.GA27731@cvs.haskell.org> Thu Mar 1 06:08:31 PST 2007 Malcolm.Wallace@cs.york.ac.uk * Use ghc rather than gcc to compile ghc_floats.c M ./src/compiler98/Makefile -1 +1 From Malcolm.Wallace at cs.york.ac.uk Thu Mar 1 09:31:30 2007 From: Malcolm.Wallace at cs.york.ac.uk (Malcolm Wallace) Date: Thu Mar 1 09:24:52 2007 Subject: [cvs-nhc98] patch applied (nhc98): hsc2hs needs to pass -m32 to the linker Message-ID: <20070301143130.GA28500@cvs.haskell.org> Thu Mar 1 06:29:27 PST 2007 Malcolm.Wallace@cs.york.ac.uk * hsc2hs needs to pass -m32 to the linker M ./script/hsc2hs.inst -1 +1 From sven.panne at aedion.de Thu Mar 1 10:47:24 2007 From: sven.panne at aedion.de (Sven Panne) Date: Thu Mar 1 10:41:44 2007 Subject: [cvs-nhc98] patch applied (nhc98): Fixed ARCH confusion In-Reply-To: <20070301135840.24e8940b.Malcolm.Wallace@cs.york.ac.uk> References: <20070301121137.GA23922@cvs.haskell.org> <20070301135840.24e8940b.Malcolm.Wallace@cs.york.ac.uk> Message-ID: <200703011647.25041.sven.panne@aedion.de> On Thursday 01 March 2007 14:58, Malcolm Wallace wrote: > [...] > > The correct way to compile ghc_floats.c would be using GHC itself, > > because then ABI compability is guaranteed. Due to my limited > > knowledge of nhc98's build system, I leave this to Malcolm... > > That would make sense. OK, with all the recent changes I get quite far, but only because I have a local modification of the nhc98 script: When processing a *.gc file, it uses "ld -r" to partially link together the C part and the Haskell part. To make this work in the presence of the -m32 hack, one has to add " -melf_i386" to the "ld -r" lines, telling ld to use the elf_i386 emulation. Other platforms will of course need other emulations. I am not sure how to fix this in a portable way, perhaps one can use "gcc -m32" somehow for this partial linking step? Hmmm, ideas would be highly welcome... Cheers, S. From Malcolm.Wallace at cs.york.ac.uk Thu Mar 1 11:15:27 2007 From: Malcolm.Wallace at cs.york.ac.uk (Malcolm Wallace) Date: Thu Mar 1 11:11:13 2007 Subject: [cvs-nhc98] patch applied (nhc98): Fixed ARCH confusion In-Reply-To: <200703011647.25041.sven.panne@aedion.de> References: <20070301121137.GA23922@cvs.haskell.org> <20070301135840.24e8940b.Malcolm.Wallace@cs.york.ac.uk> <200703011647.25041.sven.panne@aedion.de> Message-ID: <20070301161527.65b18188.Malcolm.Wallace@cs.york.ac.uk> Sven Panne wrote: > OK, with all the recent changes I get quite far, but only because I > have a local modification of the nhc98 script: When processing a *.gc > file, it uses "ld -r" to partially link together the C part and the > Haskell part. Since $LD is already defined to equal $CC, I'm thinking to replace ld -r ... with $LD -Wl,-r ... Regards, Malcolm From sven.panne at aedion.de Thu Mar 1 11:57:36 2007 From: sven.panne at aedion.de (Sven Panne) Date: Thu Mar 1 11:50:58 2007 Subject: [cvs-nhc98] patch applied (nhc98): Fixed ARCH confusion In-Reply-To: <20070301161527.65b18188.Malcolm.Wallace@cs.york.ac.uk> References: <20070301121137.GA23922@cvs.haskell.org> <200703011647.25041.sven.panne@aedion.de> <20070301161527.65b18188.Malcolm.Wallace@cs.york.ac.uk> Message-ID: <200703011757.37125.sven.panne@aedion.de> On Thursday 01 March 2007 17:15, Malcolm Wallace wrote: > Since $LD is already defined to equal $CC, I'm thinking to replace > ld -r ... > with > $LD -Wl,-r ... I've just experimented a bit, but only with partial success: When gcc is called with the -m32 option, it passes the correct emulation option to the linker, so this part works. What doesn't work: gcc is not aware that we only want to do a partial linking, despite of the -Wl,r option, and tries to link things like -lc, -lgcc, ... :-( Is this partial linking really necessary? Or can the driver script be adapted to handle 2 .o files explicitly when dealing with *.gc files? This might be a cleaner solution. Cheers, S. From Malcolm.Wallace at cs.york.ac.uk Thu Mar 1 12:33:13 2007 From: Malcolm.Wallace at cs.york.ac.uk (Malcolm Wallace) Date: Thu Mar 1 12:30:50 2007 Subject: [cvs-nhc98] patch applied (nhc98): Fixed ARCH confusion In-Reply-To: <200703011757.37125.sven.panne@aedion.de> References: <20070301121137.GA23922@cvs.haskell.org> <200703011647.25041.sven.panne@aedion.de> <20070301161527.65b18188.Malcolm.Wallace@cs.york.ac.uk> <200703011757.37125.sven.panne@aedion.de> Message-ID: <20070301173313.6ed7da03.Malcolm.Wallace@cs.york.ac.uk> Sven Panne wrote: > On Thursday 01 March 2007 17:15, Malcolm Wallace wrote: > > Since $LD is already defined to equal $CC, I'm thinking to replace > > ld -r ... > > with > > $LD -Wl,-r ... > > What doesn't work: gcc is > not aware that we only want to do a partial linking, despite of the > -Wl,r option, and tries to link things like -lc, -lgcc, ... :-( Hmm. How about telling gcc -nostdlib and/or -nodefaultlibs? > Is this partial linking really necessary? Or can the driver script be > adapted to handle 2 .o files explicitly when dealing with *.gc files? I can't remember exactly why the partial linking was introduced, but have a feeling it was to make things cleaner, not dirtier. :-) Possibly because hmake only needs to look for one object file. Or possibly because the ghc version of greencard only generated a single object. Regards, Malcolm From Sven.Panne at aedion.de Fri Mar 2 08:39:49 2007 From: Sven.Panne at aedion.de (Sven Panne) Date: Fri Mar 2 08:33:08 2007 Subject: [cvs-nhc98] patch applied (nhc98): Use GCC for partial linking Message-ID: <20070302133949.GA28138@cvs.haskell.org> Fri Mar 2 05:33:35 PST 2007 sven.panne@aedion.de * Use GCC for partial linking To use the correct emulation, use GCC for partial linking, passing it the -r linker option and telling it to not link any standard libs into the resulting object file. M ./script/nhc98.inst -2 +2 From Malcolm.Wallace at cs.york.ac.uk Fri Mar 2 08:58:56 2007 From: Malcolm.Wallace at cs.york.ac.uk (Malcolm Wallace) Date: Fri Mar 2 08:52:15 2007 Subject: [cvs-nhc98] patch applied (nhc98): Prelude sum and product should really use foldl' not foldl. Message-ID: <20070302135856.GA28763@cvs.haskell.org> Thu Mar 1 08:54:37 PST 2007 Malcolm.Wallace@cs.york.ac.uk * Prelude sum and product should really use foldl' not foldl. M ./src/prelude/PreludeList/Foldl.hs +4 M ./src/prelude/PreludeList/Product.hs -1 +3 M ./src/prelude/PreludeList/Sum.hs -1 +3 From Sven.Panne at aedion.de Fri Mar 2 09:34:35 2007 From: Sven.Panne at aedion.de (Sven Panne) Date: Fri Mar 2 09:27:54 2007 Subject: [cvs-nhc98] patch applied (nhc98): Synch spec file with reality Message-ID: <20070302143435.GA30399@cvs.haskell.org> Fri Mar 2 06:33:37 PST 2007 sven.panne@aedion.de * Synch spec file with reality M ./nhc98.spec -7 +9 From igloo at earth.li Fri Mar 2 09:51:09 2007 From: igloo at earth.li (Ian Lynagh) Date: Fri Mar 2 09:44:28 2007 Subject: [cvs-nhc98] patch applied (nhc98): Prelude sum and product should really use foldl' not foldl. In-Reply-To: <20070302135856.GA28763@cvs.haskell.org> References: <20070302135856.GA28763@cvs.haskell.org> Message-ID: <20070302145109.GA10235@matrix.chaos.earth.li> On Fri, Mar 02, 2007 at 05:58:56AM -0800, Malcolm Wallace wrote: > Thu Mar 1 08:54:37 PST 2007 Malcolm.Wallace@cs.york.ac.uk > * Prelude sum and product should really use foldl' not foldl. Not according to the Haskell 98 report: http://haskell.org/onlinereport/standard-prelude.html#$vsum The difference is shown up by: ------------------------------------------------- import Data.List data Foo = Zero | A | B | C deriving (Eq, Show) instance Num Foo where _ + Zero = error "Adding Zero" _ + A = error "Adding A" _ + B = error "Adding B" _ + C = C fromInteger 0 = Zero xs = [A, B, C] f = foldl (+) 0 xs f' = foldl' (+) 0 xs s = sum xs ------------------------------------------------- ghc (6.6 and HEAD) agrees with the report: *Main> f C *Main> f' *** Exception: Adding A *Main> s C *Main> while hugs (September 2006) disagrees: Main> f C Main> f' Program error: Adding A Main> s Program error: Adding A Main> I guess a decision should be made for Haskell', probably either going with foldl' or introducing sum' and product'. Thanks Ian From Malcolm.Wallace at cs.york.ac.uk Fri Mar 2 10:48:00 2007 From: Malcolm.Wallace at cs.york.ac.uk (Malcolm Wallace) Date: Fri Mar 2 10:50:09 2007 Subject: [cvs-nhc98] patch applied (nhc98): Prelude sum and product should really use foldl' not foldl. In-Reply-To: <20070302145109.GA10235@matrix.chaos.earth.li> References: <20070302135856.GA28763@cvs.haskell.org> <20070302145109.GA10235@matrix.chaos.earth.li> Message-ID: <20070302154800.4ff56caa.Malcolm.Wallace@cs.york.ac.uk> Ian Lynagh wrote: > On Fri, Mar 02, 2007 at 05:58:56AM -0800, Malcolm Wallace wrote: > > Thu Mar 1 08:54:37 PST 2007 Malcolm.Wallace@cs.york.ac.uk > > * Prelude sum and product should really use foldl' not foldl. > > Not according to the Haskell 98 report: OK, you are the second person to complain within an hour of the checkin! Maybe I should revert. I was just trying to fix an egregious space leak for the standard numeric types. There are not too many genuinely lazy numeric datatypes around. Regards, Malcolm From isaacdupree at charter.net Fri Mar 2 16:38:03 2007 From: isaacdupree at charter.net (Isaac Dupree) Date: Fri Mar 2 16:30:24 2007 Subject: [cvs-nhc98] segfault-causing examples Message-ID: <45E8993B.9020709@charter.net> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 main = do print "hello" print (True) print "bye" compiling with my darcs-nhc98 on my computer and running, I get: "hello" Segmentation fault (core dumped) Before trying that, I noticed that Integers had some problems... module Main(main) where main = do print "hello" if 2 > (1::Int) then print "SmartInt" else print "StupidInt" if 2 > (1::Integer) then print "SmartIntegerA" else print "StupidIntegerA" if 1 < (2::Integer) then print "SmartIntegerB" else print "StupidIntegerB" print (1::Int) print (1::Integer) print "bye" My results, plus comments verified by other experiments: "hello" "SmartInt" --Ints are fine "StupidIntegerA" --Integers are "SmartIntegerB" --not working right 1 --Ints are still fine Segmentation fault (core dumped) --Instance Show Integer crashes the program show (1::Integer) `seq` print "integ" --prints "integ" (shallow seq) foldr seq () (show (1::Integer)) `seq` print "integ" --crashes (deep seq) I suppose maybe I compiled nhc wrong, since I manually modified an "==" to "=" in one file, and restarted its compilation several times, though I can't see how those could cause these problems... Isaac -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.3 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFF6Jk6HgcxvIWYTTURApy5AJ971v1rRcy6fqnBrTheaQarakIPNgCfXZmg sRq1AhmrIxPIW9bY58bbbUw= =FLqA -----END PGP SIGNATURE----- From Malcolm.Wallace at cs.york.ac.uk Fri Mar 2 17:54:29 2007 From: Malcolm.Wallace at cs.york.ac.uk (Malcolm Wallace) Date: Fri Mar 2 17:47:30 2007 Subject: [cvs-nhc98] segfault-causing examples In-Reply-To: <45E8993B.9020709@charter.net> References: <45E8993B.9020709@charter.net> Message-ID: <20070302225429.433eb473.Malcolm.Wallace@cs.york.ac.uk> Isaac Dupree writes: > main = do > if 1 < (2::Integer) then print "SmartIntegerB" else print > "StupidIntegerB" > > "SmartIntegerB" --not working right That looks like the correct result to me. > Segmentation fault (core dumped) --Instance Show Integer crashes the program Interesting. > I suppose maybe I compiled nhc wrong, since I manually modified an "==" > to "=" in one file, and restarted its compilation several times, though > I can't see how those could cause these problems... The "==" to "=" change was fine, and there is now a patch in the repo for that anyway. You say that your gcc gives random crashes. My best guess is flaky hardware - bad memory. Maybe there was some random corruption in gcc whilst building something in the runtime system. I would try rm targets/powerpc-Darwin8/obj/runtime rm targets/powerpc-Darwin8/runtime make runtime and see if you get the same behaviour. Regards, Malcolm From isaacdupree at charter.net Fri Mar 2 18:31:06 2007 From: isaacdupree at charter.net (Isaac Dupree) Date: Fri Mar 2 18:23:23 2007 Subject: [cvs-nhc98] segfault-causing examples In-Reply-To: <20070302225429.433eb473.Malcolm.Wallace@cs.york.ac.uk> References: <45E8993B.9020709@charter.net> <20070302225429.433eb473.Malcolm.Wallace@cs.york.ac.uk> Message-ID: <45E8B3BA.5020201@charter.net> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Malcolm Wallace wrote: > You say that your gcc gives random crashes. My best > guess is flaky hardware - bad memory. Yes, I know that's the usual theory, I've run memory tests and haven't detected any. Maybe gcc is buggy and corrupts its stack unpredictably. Although that could just as well produce incorrect results sometimes, I suppose. As a reference point, my machine is a Yhc buildbot and that seems to work fine when gcc (called by ghc) doesn't outright crash. Hmm... compiling with -O3 with "gcc (GCC) 4.1.2 20060928 (prerelease) (Ubuntu 4.1.1-13ubuntu5)" things (including bits of GMP) maybe isn't quite safe? Apparently C compilers are unreliable enough these days that "It has now come to the point where a compiler should be assumed to miscompile GMP." and "It might seem that GMP is unusually unlucky with compilers, that it triggers more bugs than other packages. I believe that is an illusion. What really is happening, is that GMP has a test suite that will find almost every miscompilation in the library." > Maybe there was some random > corruption in gcc whilst building something in the runtime system. I > would try > rm targets/powerpc-Darwin8/obj/runtime > rm targets/powerpc-Darwin8/runtime > make runtime > and see if you get the same behaviour. rm -r targets/ppc-Linux/obj/runtime/ rm targets/ppc-Linux/runtime make runtime make install This didn't help. BTW my --prefix=/home/isaac/install/nhc98 (and I haven't put that/bin in my PATH) Isaac -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.3 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFF6LO5HgcxvIWYTTURAnnvAKCB1Oq3irRKx8ZyTEibfpmoqWSYmwCgvIqK P1WL7QS06iBHkdCMp5s72Fk= =5wWy -----END PGP SIGNATURE----- From Malcolm.Wallace at cs.york.ac.uk Sat Mar 3 13:18:14 2007 From: Malcolm.Wallace at cs.york.ac.uk (Malcolm Wallace) Date: Sat Mar 3 13:11:30 2007 Subject: [cvs-nhc98] patch applied (nhc98): Prelude sum and product should really use foldl' not foldl. Message-ID: <20070303181814.GA7528@cvs.haskell.org> Thu Mar 1 08:54:37 PST 2007 Malcolm.Wallace@cs.york.ac.uk UNDO: Prelude sum and product should really use foldl' not foldl. M ./src/prelude/PreludeList/Foldl.hs -4 M ./src/prelude/PreludeList/Product.hs -3 +1 M ./src/prelude/PreludeList/Sum.hs -3 +1 From sven.panne at aedion.de Sun Mar 4 08:04:38 2007 From: sven.panne at aedion.de (Sven Panne) Date: Sun Mar 4 07:57:49 2007 Subject: [cvs-nhc98] Building source distributions Message-ID: <200703041404.38463.sven.panne@aedion.de> I'm having trouble building source distributions: After the ./configure stage, "make hmakeDist" seems to work, but the "srcDist" target has problems. First of all, it seems to depend on "basic", but this is not recorded in the Makefile. After fixing that and a looong time, "make srcDist" fails with: ================================================= ... make[1]: Entering directory `/home/panne/nhc98/src/interpreter' rm -rf *.o /home/panne/nhc98/targets/x86_64-Linux/obj/interpreter rm -f *.hi rm -f *.hc *.c /home/panne/nhc98/script/hmake -hc=/home/panne/nhc98/script/nhc98 -I../hmake -C HInteractive.hs /home/panne/nhc98/script/nhc98 -C -I../hmake -package base -package readline -DUSE_READLINE=1 -lreadline -c -cpp -o LexModule.hc LexModule.hs Warning: -package readline interfaces not found. Warning: -package readline library not found. /home/panne/nhc98/script/nhc98 -C -I../hmake -package base -package readline -DUSE_READLINE=1 -lreadline -c -cpp -o SimpleLineEditor.hc SimpleLineEditor.hs Warning: -package readline interfaces not found. nhc98comp: user error (Can't open any of: ./System/Console/Readline.hi ../hmake/System/Console/Readline.hi /home/panne/nhc98/include/packages/base/System/Console/Readline.hi /home/panne/nhc98/include/System/Console/Readline.hi when trying to read System.Console.Readline. ) make[1]: *** [cfiles] Fehler 1 make[1]: Leaving directory `/home/panne/nhc98/src/interpreter' make: *** [targets/hmakeC] Fehler 2 ================================================= Am I doing something wrong? Or is srcDist currently simply broken? Cheers, S. From Malcolm.Wallace at cs.york.ac.uk Mon Mar 5 05:55:53 2007 From: Malcolm.Wallace at cs.york.ac.uk (Malcolm Wallace) Date: Mon Mar 5 05:50:31 2007 Subject: [cvs-nhc98] segfault-causing examples In-Reply-To: <45E8B3BA.5020201@charter.net> References: <45E8993B.9020709@charter.net> <20070302225429.433eb473.Malcolm.Wallace@cs.york.ac.uk> <45E8B3BA.5020201@charter.net> Message-ID: <20070305105553.1d71c314.Malcolm.Wallace@cs.york.ac.uk> Isaac Dupree wrote: > > Maybe there was some random > > corruption in gcc whilst building something in the runtime system. > > rm -r targets/ppc-Linux/obj/runtime/ > rm targets/ppc-Linux/runtime > make runtime > make install > > This didn't help. I'm afraid I'm out of ideas then. Is your powerpc a 64-bit G5, or a 32-bit G3/G4? If a 64-bit machine, does your version of gcc also generate 64-bit code by default, or it is set to 32-bit mode? Regards, Malcolm From Malcolm.Wallace at cs.york.ac.uk Mon Mar 5 06:03:47 2007 From: Malcolm.Wallace at cs.york.ac.uk (Malcolm Wallace) Date: Mon Mar 5 06:04:28 2007 Subject: [cvs-nhc98] Building source distributions In-Reply-To: <200703041404.38463.sven.panne@aedion.de> References: <200703041404.38463.sven.panne@aedion.de> Message-ID: <20070305110347.061f9a64.Malcolm.Wallace@cs.york.ac.uk> Sven Panne wrote: > I'm having trouble building source distributions: > ... the "srcDist" target ... seems to depend on "basic", but > this is not recorded in the Makefile. Yeah. The srcDist target was never really intended to be stand-alone. It does require that the build tree should already contain a built-and-working compiler/hmake/cpphs in order to be able to create the bootstrapping .hc files. (And yes, it would probably be a good idea to record this explicitly in the Makefile.) > /home/panne/nhc98/script/hmake -hc=/home/panne/nhc98/script/nhc98 > -I../hmake -C HInteractive.hs > /home/panne/nhc98/script/nhc98 -C -I../hmake -package base > -package readline -DUSE_READLINE=1 -lreadline -c -cpp -o LexModule.hc > LexModule.hs Warning: -package readline interfaces not found. > > Am I doing something wrong? Or is srcDist currently simply broken? This part is broken at the moment, yes. If you built hmake/interpreter with ghc originally, then srcDist attempts to create bootstrapping .hc files using the ghc-based configuration (in particular, for readline), even though nhc98 does not know about the readline package. One solution would be to build readline with nhc98. Another might be to disable using/building the bootstrapping .hc files for src/interpreter. Regards, Malcolm From Malcolm.Wallace at cs.york.ac.uk Mon Mar 5 15:56:00 2007 From: Malcolm.Wallace at cs.york.ac.uk (Malcolm Wallace) Date: Mon Mar 5 15:49:11 2007 Subject: [cvs-nhc98] patch applied (hat): pass BUILDOPTS to C compilation as well as Haskell Message-ID: <20070305205600.GA843@cvs.haskell.org> Mon Mar 5 01:29:07 PST 2007 Malcolm.Wallace@cs.york.ac.uk * pass BUILDOPTS to C compilation as well as Haskell M ./src/hattools/Makefile -2 +2 From Malcolm.Wallace at cs.york.ac.uk Tue Mar 6 05:21:35 2007 From: Malcolm.Wallace at cs.york.ac.uk (Malcolm Wallace) Date: Tue Mar 6 05:14:44 2007 Subject: [cvs-nhc98] patch applied (hat): comments only Message-ID: <20070306102135.GA25812@cvs.haskell.org> Wed Feb 7 08:02:02 PST 2007 Malcolm.Wallace@cs.york.ac.uk * comments only M ./src/hattrans/Flags.hs -2 +2 From Malcolm.Wallace at cs.york.ac.uk Tue Mar 6 05:21:37 2007 From: Malcolm.Wallace at cs.york.ac.uk (Malcolm Wallace) Date: Tue Mar 6 05:14:45 2007 Subject: [cvs-nhc98] patch applied (hat): fix quoting of BUILDOPTS in hat-tools Makefile Message-ID: <20070306102137.GA25833@cvs.haskell.org> Tue Mar 6 02:07:29 PST 2007 Malcolm.Wallace@cs.york.ac.uk * fix quoting of BUILDOPTS in hat-tools Makefile M ./src/hattools/Makefile -5 +7 From Malcolm.Wallace at cs.york.ac.uk Wed Mar 7 05:37:25 2007 From: Malcolm.Wallace at cs.york.ac.uk (Malcolm Wallace) Date: Wed Mar 7 05:30:28 2007 Subject: [cvs-nhc98] patch applied (hat): typo Message-ID: <20070307103725.GA18355@cvs.haskell.org> Wed Mar 7 02:35:29 PST 2007 Malcolm.Wallace@cs.york.ac.uk * typo M ./src/hattools/Makefile -1 +1 From Malcolm.Wallace at cs.york.ac.uk Wed Mar 7 05:37:26 2007 From: Malcolm.Wallace at cs.york.ac.uk (Malcolm Wallace) Date: Wed Mar 7 05:30:30 2007 Subject: [cvs-nhc98] patch applied (hat): BUILDOPTS required for C compilation in hatlib also Message-ID: <20070307103726.GA18375@cvs.haskell.org> Wed Mar 7 02:36:35 PST 2007 Malcolm.Wallace@cs.york.ac.uk * BUILDOPTS required for C compilation in hatlib also M ./src/hatlib/Makefile -1 +1 From isaacdupree at charter.net Wed Mar 7 19:46:58 2007 From: isaacdupree at charter.net (Isaac Dupree) Date: Wed Mar 7 19:39:01 2007 Subject: [cvs-nhc98] segfault-causing examples In-Reply-To: <20070305105553.1d71c314.Malcolm.Wallace@cs.york.ac.uk> References: <45E8993B.9020709@charter.net> <20070302225429.433eb473.Malcolm.Wallace@cs.york.ac.uk> <45E8B3BA.5020201@charter.net> <20070305105553.1d71c314.Malcolm.Wallace@cs.york.ac.uk> Message-ID: <45EF5D02.50905@charter.net> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 I tried completely rebuilding from a clean checkout, first changing -O3 to -O0 in Makefile.inc. It didn't help. Malcolm Wallace wrote: > I'm afraid I'm out of ideas then. Is your powerpc a 64-bit G5, or a > 32-bit G3/G4? If a 64-bit machine, does your version of gcc also > generate 64-bit code by default, or it is set to 32-bit mode? It's a 32-bit G3. Isaac -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.3 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFF710BHgcxvIWYTTURAosoAJ9x4NpYEPBYkdXpdGKdVTENcHdD9wCg2C+z KiS+j1rMCX2nRLCooJRPPTs= =ioXH -----END PGP SIGNATURE----- From Malcolm.Wallace at cs.york.ac.uk Fri Mar 9 15:51:35 2007 From: Malcolm.Wallace at cs.york.ac.uk (Malcolm Wallace) Date: Fri Mar 9 15:44:33 2007 Subject: [cvs-nhc98] patch applied (hat): comment out all installation-time references to hier lib dirs Message-ID: <20070309205135.GA17908@cvs.haskell.org> Fri Mar 9 12:49:57 PST 2007 Malcolm.Wallace@cs.york.ac.uk * comment out all installation-time references to hier lib dirs M ./configure -22 +22 M ./script/confhat -3 +3 M ./src/hatlib/Makefile -6 +6 From Malcolm.Wallace at cs.york.ac.uk Mon Mar 12 06:17:48 2007 From: Malcolm.Wallace at cs.york.ac.uk (Malcolm Wallace) Date: Mon Mar 12 06:10:35 2007 Subject: [cvs-nhc98] patch applied (hat): fix warnings about malloc Message-ID: <20070312101748.GA31101@cvs.haskell.org> Fri Mar 9 08:44:43 PST 2007 Malcolm.Wallace@cs.york.ac.uk * fix warnings about malloc M ./src/hattools/nontermutils.c +1 M ./src/hattools/observeutils.c +1 M ./src/hattools/parentset.c +1 From Malcolm.Wallace at cs.york.ac.uk Mon Mar 12 06:17:49 2007 From: Malcolm.Wallace at cs.york.ac.uk (Malcolm Wallace) Date: Mon Mar 12 06:10:37 2007 Subject: [cvs-nhc98] patch applied (hat): Cabal comment format should be -- not # Message-ID: <20070312101749.GA31122@cvs.haskell.org> Mon Mar 12 03:13:46 PDT 2007 Malcolm.Wallace@cs.york.ac.uk * Cabal comment format should be -- not # M ./src/hatlib/hat.cabal -65 +65 From Malcolm.Wallace at cs.york.ac.uk Tue Mar 13 06:41:43 2007 From: Malcolm.Wallace at cs.york.ac.uk (Malcolm Wallace) Date: Tue Mar 13 06:41:43 2007 Subject: [cvs-nhc98] patch applied (hat): fix harch on recent Linux to report ix86 rather than e.g. athlon-4 Message-ID: <20070313104143.GA14564@cvs.haskell.org> Tue Mar 13 03:40:35 PDT 2007 Malcolm.Wallace@cs.york.ac.uk * fix harch on recent Linux to report ix86 rather than e.g. athlon-4 M ./script/harch -1 +1 From Malcolm.Wallace at cs.york.ac.uk Fri Mar 16 12:28:03 2007 From: Malcolm.Wallace at cs.york.ac.uk (Malcolm Wallace) Date: Fri Mar 16 12:27:53 2007 Subject: [cvs-nhc98] patch applied (nhc98): add new package filepath to default libraries build Message-ID: <20070316162803.GA32224@cvs.haskell.org> Fri Mar 16 09:18:33 PDT 2007 Malcolm.Wallace@cs.york.ac.uk * add new package filepath to default libraries build M ./Makefile -1 +1 M ./script/pkgdirlist +3 M ./src/libraries/default-packages +1 From Malcolm.Wallace at cs.york.ac.uk Tue Mar 20 08:12:48 2007 From: Malcolm.Wallace at cs.york.ac.uk (Malcolm Wallace) Date: Tue Mar 20 08:12:36 2007 Subject: [cvs-nhc98] patch applied (nhc98): Library packages no longer need a Makefile.inc in their top directory. Message-ID: <20070320121248.GA31649@cvs.haskell.org> Tue Mar 20 04:59:33 PDT 2007 Malcolm.Wallace@cs.york.ac.uk * Library packages no longer need a Makefile.inc in their top directory. M ./src/libraries/Makefile.inc -1 +1 From Malcolm.Wallace at cs.york.ac.uk Tue Mar 20 16:17:23 2007 From: Malcolm.Wallace at cs.york.ac.uk (Malcolm Wallace) Date: Tue Mar 20 16:16:59 2007 Subject: [cvs-nhc98] patch applied (nhc98): Teach the evil mangler about apple/x86 assembler alignment. Message-ID: <20070320201723.GA15435@cvs.haskell.org> Tue Mar 20 13:10:28 PDT 2007 Malcolm.Wallace@cs.york.ac.uk * Teach the evil mangler about apple/x86 assembler alignment. This patch enables nhc98 to build/work correctly on Apple/Intel machines. M ./script/nhc98.inst +1 From sven.panne at aedion.de Fri Mar 30 08:40:37 2007 From: sven.panne at aedion.de (Sven Panne) Date: Fri Mar 30 08:39:42 2007 Subject: [cvs-nhc98] patch applied (nhc98): add new package filepath to default libraries build In-Reply-To: <20070316162803.GA32224@cvs.haskell.org> References: <20070316162803.GA32224@cvs.haskell.org> Message-ID: <200703301440.37355.sven.panne@aedion.de> On Friday 16 March 2007 17:28, Malcolm Wallace wrote: > Fri Mar 16 09:18:33 PDT 2007 Malcolm.Wallace@cs.york.ac.uk > * add new package filepath to default libraries build > > M ./Makefile -1 +1 > M ./script/pkgdirlist +3 > M ./src/libraries/default-packages +1 GHC uses a different version than nhc98 now. I guess that nhc98 should get this package from darcs.haskell.org, too. Or is there a reason not to do that? Cheers, S. From Malcolm.Wallace at cs.york.ac.uk Fri Mar 30 09:31:16 2007 From: Malcolm.Wallace at cs.york.ac.uk (Malcolm Wallace) Date: Fri Mar 30 09:30:21 2007 Subject: [cvs-nhc98] patch applied (nhc98): change the default location of the filepath repo Message-ID: <20070330133116.GA6964@cvs.haskell.org> Fri Mar 30 06:30:21 PDT 2007 Malcolm.Wallace@cs.york.ac.uk * change the default location of the filepath repo M ./src/libraries/default-packages -1 +1