Hello, On PPC Max OS X in a recent HEAD, pp1 fails as follows:
> =====> pp1(normal) > cd ./lib/PrettyPrint && '/Users/thorkilnaur/tn/buildbot/ghc/tnaur-ppc-osx/tnaur-ppc-osx-head/build/compiler/stage2/ghc-inplace' -no-recomp -dcore-lint -dcmm-lint -Dpowerpc_apple_darwin -o pp1 pp1.hs >pp1.comp.stderr 2>&1 > Compile failed (status 256) errors were: > /usr/bin/ld: Undefined symbols: > ___stginit_prettyzm1zi0_TextziPrettyPrintziHughesPJ_ > _prettyzm1zi0_TextziPrettyPrintziHughesPJ_cat_closure > _prettyzm1zi0_TextziPrettyPrintziHughesPJ_char_closure > _prettyzm1zi0_TextziPrettyPrintziHughesPJ_nest_closure > _prettyzm1zi0_TextziPrettyPrintziHughesPJ_parens_closure > _prettyzm1zi0_TextziPrettyPrintziHughesPJ_sep_closure > _prettyzm1zi0_TextziPrettyPrintziHughesPJ_text_closure > _prettyzm1zi0_TextziPrettyPrintziHughesPJ_zdf1_closure > collect2: ld returned 1 exit status The tnaur-PPX-OSX-head builder has this failure mode for the first time for "test run at Fri Jun 8 21:45:39 CEST 2007". The expected failure looks like this, last seen for "test run at Mon May 7 21:50:19 CEST 2007": > =====> pp1(normal) > cd ./lib/PrettyPrint && '/Users/thorkilnaur/tn/buildbot/ghc/tnaur-ppc-osx/tnaur-ppc-osx-head/build/compiler/stage2/ghc-inplace' -no-recomp -dcore-lint -dcmm-lint -Dpowerpc_apple_darwin -o pp1 pp1.hs >pp1.comp.stderr 2>&1 > cd ./lib/PrettyPrint && ./pp1 </dev/null >pp1.run.stdout 2>pp1.run.stderr > Actual stdout output differs from expected: > *** ./lib/PrettyPrint/pp1.stdout Wed Mar 14 19:15:05 2007 > --- ./lib/PrettyPrint/pp1.run.stdout Thu Mar 15 02:00:40 2007 > *************** > *** 1,4 **** > ! This output is not what is expected, becuase the > ! test "works" now, by virtue of a hack in HughesPJ.spaces. > ! I'm leaving this strange output here to remind us to look > ! at the root cause of the problem. Sometime. > \ No newline at end of file > --- 1,51 ---- > ! (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa > ! + > > a > ! a > ! a > ... > ! a) I can reproduce this by hand and also correct it by adding a -package pretty option. I presume this is related to the base package split effort. But I am not able to tell just how to proceed with this, so help would be greatly appreciated. Thanks and best regards Thorkil _______________________________________________ Cvs-ghc mailing list [email protected] http://www.haskell.org/mailman/listinfo/cvs-ghc
