https://gcc.gnu.org/bugzilla/show_bug.cgi?id=123976
--- Comment #244 from Sergey Fedorov <vital.had at gmail dot com> --- (In reply to Peter Dyballa from comment #233) > > > > It is possible that one of the system tools has a bug that's not manifested > > in many previous versions of libgcc, of course - but Occam's razor suggests > > it's more likely some simpler explanation - like an unexpected change. > > And my assumption is that this change is somewhere in building the Makefile > rules for PPC because it seems obvious that on more powerful i686, PPC64 and > maybe x86_64 it works. (For me Rosetta is special, a fifth mixed case. But I > could be clinging too much to my own findings. Therefore I have made > distclean my PowerBook G4 and am compiling and compiling and commenting …) The build works fine, reproducibly, from my portfile of gcc16 on ppc on 10.6.8. Yes, it is a "non-standard" system in a sense it is not an official release from Apple, and it does not automatically translates into "this must work on 10.5 too", however it still illustrates that the issue cannot be reduced to "this works everywhere else but broken on 32-bit ppc". P. S. I repeat myself, but since the topic is huge by now, let's avoid making statements which may be interpreted in a wrong way, that does not make debugging easier.
