Building GHC on OS X

../compiler/ghc-inplace -H32m -O0 -W -fno-warn-unused-matches - fwarn-unused-imports -istage2/utils -istage2/basicTypes -istage2/ types -istage2/hsSyn -istage2/prelude -istage2/rename -istage2/ typecheck -istage2/deSugar -istage2/coreSyn -istage2/specialise - istage2/simplCore -istage2/stranal -istage2/stgSyn -istage2/ simplStg -istage2/codeGen -istage2/main -istage2/profiling - istage2/parser -istage2/cprAnalysis -istage2/ndpFlatten -istage2/ iface -istage2/cmm -istage2/nativeGen -istage2/ghci -Istage2 - DGHCI -DBREAKPOINT -package template-haskell -threaded -cpp -fglasgow- exts -fno-generics -Rghc-timing -I. -IcodeGen -InativeGen -Iparser -

Hi, I've been using ghc on Mac OS X for a while now, so when I noticed Simon Marlow's call (made last month, but I just noticed it) for people to help with OS X, I thought, why not? Well, it turns out that I can't even build ghc, so I'm not likely to end up being much help. But it would be nice to at least be able to build the thing, so if anybody can help me... I'm trying to build GHC (6.5.?? ... version from darcs.haskell.org as of yesterday) on Mac OS X (10.4.7 ppc) bootstrapping from ghc 6.4.2 and gcc 4.0.1. "./configure; make" does not work. On my initial try, I ran into the UNDO_DELETE, etc problems associated with Darwin's partial readline support described in ticket #766. The fix suggested in that ticket seems to work fine, but not until I installed the GNU readline library. (If the change to detect the broken Darwin readline is in the repository, it did not work for me). After fixing the readline problem, it looks like I can build at least part of the stage1 compiler -- but the build fails just after starting to build stage2. The final error is: package unix -package Cabal -ignore-package lang -recomp -Rghc-timing -DDEBUG -H16M '-#include "hschooks.h"' -package-name ghc-6.5 - fgenerics -O -c utils/FastString.lhs -o stage2/utils/FastString.o - ohi stage2/utils/FastString.hi

utils/FastString.lhs:73:0: Failed to load interface for `GHC.Exts': Could not find module `GHC.Exts': use -v to see a list of the files searched for <

(1 samples), 16M in use, 0.00 INIT (0.00 elapsed), 0.29 MUT (0.38 elapsed), 0.03 GC (0.04 elapsed) :ghc>>

make[2]: *** [stage2/utils/FastString.o] Error 1 make[1]: *** [stage2] Error 2 make: *** [bootstrap2] Error 2

The first error that I can find comes when building ghc/libraries/ base (in stage1)

../../compiler/ghc-inplace -H32m -O0 -W -fno-warn-unused-matches - fwarn-unused-imports -fglasgow-exts -cpp -Iinclude -"#include" HsBase.h -funbox-strict-fields -package-name base-1.0 -O -dcore-lint -W -fno-warn-unused-matches -fwarn-unused-imports -keep-hc-files - fgenerics -c Data/Generics/Schemes.hs -o Data/Generics/Schemes.o - ohi Data/Generics/Schemes.hi

Data/Generics/Schemes.hs:108:15: Couldn't match expected type `r' (a rigid variable) against inferred type `[a]' `r' is bound by the type signature for `listify' at Data/Generics/Schemes.hs:106:20 Expected type: r -> r -> r Inferred type: [a] -> [a] -> [a] In the first argument of `everything', namely `(++)' In the call (everything (++) ([] `mkQ` (\ x -> if p x then [x] else []))) make[2]: *** [Data/Generics/Schemes.o] Error 1

After that, there are a handful of errors complaining about things that can't be found (e.g.)

CPUTime.hs:5:0: Failed to load interface for `System.CPUTime': Could not find module `System.CPUTime': use -v to see a list of the files searched for make[2]: *** [CPUTime.o] Error 1

and

Language/Haskell/TH/Syntax.hs:46:0: Failed to load interface for `Data.PackedString': Could not find module `Data.PackedString': use -v to see a list of the files searched for make[2]: *** [Language/Haskell/TH/Syntax.o] Error 1

There are six or seven more messages like those, but nothing severe enough to stop make until the problem where it can not find 'GHC.Exts'. I tried to brute force it with 'make -k', which went a lot further, but in the end build nothing useful. Anybody have any ideas? Or build instructions for OS X? Thanks, -david k.