Re: configuration failed.

Malcolm> In this case I think that the real problem is that you're Malcolm> passing "-b i486-linuxaout" to the test. I don't think Malcolm> that there are many X86 cross-compilers found on PPC Malcolm> machines :-) You may be right, however the evidence is very strong that that isn't the problem (gcc is known not to handle -z, the error message mentions nocombreloc, it was executing cc1 not ld, and --pdump worked). It may be _a_ problem, and I'll take a look at the configure code, though. We should probably get rid of that -b flag. Stephen> Probably we should force use of ld directly on that Stephen> platform. "Probably." I really don't understand why we currently insist on using gcc as a linker instead of ld itself, as we go to some trouble to find out what ld is anyway. Until I understand more, I'm not going to do anything that doesn't fix a manifest bug. However, I've already applied one patch that _does_ fix a bug that manifests because gcc DTWT in some cases. Malcolm> I don't think that's necessary. I've has no problems Malcolm> building bleeding edge XEmacs with "configure Malcolm> --prefix=/usr/local/gcc3-world --with-gnome" on a Mac for Malcolm> months using gcc 2.95 and 3.0.4. And we don't use any -z flags except nocombreloc on Linux and ignore on Solaris. If we ever do use a -z flag on Linux/PPC, we'll get the same problem again. No? Simple test: configure with LDFLAGS='-z nosuchkeyword' in the environment. Betcha configure don't like that much. -- Institute of Policy and Planning Sciences http://turnbull.sk.tsukuba.ac.jp University of Tsukuba Tennodai 1-1-1 Tsukuba 305-8573 JAPAN Don't ask how you can "do" free software business; ask what your business can "do for" free software.