https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117110
--- Comment #5 from Iain Sandoe <iains at gcc dot gnu.org> --- (In reply to Jakub Jelinek from comment #4) > I'm leaning towards reversion of the r15-4290 and partial reversion of > r15-4286 (Makefile.in bits and some genmatch.cc bits) and reimplementing > whatever libcpp needs in genmatch.cc (diagnostic_cb) or its new helper, so > partial copy of pretty-print.cc printf handling with all dependencies on > diagnostic/pretty-print removed. The alternative might be to have a $build version of libcommon.a in gcc/build (it seems too complex to arrange for it to be built from the top-level into build-triple) .. I was looking at how complex it might be this morning - but have to shift to doing some other things... > Because I'm afraid as I did it, I'm linking the build genmatch with host > libcommon.a, and even if it didn't make problems on OSes without proper > intl/iconv stuff in libc, I think (though haven't tried) it breaks canadian > crosses. That seems so - and for $build!=$host==$target too.
