I've conducted some tests building glibc-2.2.3+nis patch with gcc-2.95-4 from Debian unstable and some local builds of gcc-3.0. After reading some archives of the glibc mailing lists, I discovered some oddities on the build process.
glibc uses `gcc -print-file-name=cpp` while building parts of the sunrpc subdir. (eg : /usr/local/src/build/libc/elf/ld-linux.so.2 --library-path /usr/local/src/build/libc:/usr/local/src/build/libc/math:/usr/local/src/build/libc/elf:/usr/local/src/build/libc/dlfcn:/usr/local/src/build/libc/nss:/usr/local/src/build/libc/nis:/usr/local/src/build/libc/rt:/usr/local/src/build/libc/resolv:/usr/local/src/build/libc/crypt:/usr/local/src/build/libc/linuxthreads /usr/local/src/build/libc/sunrpc/rpcgen -Y `gcc -print-file-name=cpp | sed 's|/cpp$||'` -c rpcsvc/bootparam_prot.x -o /usr/local/src/build/libc/sunrpc/xbootparam_prot.T /bin/sh ../scripts/move-if-change /usr/local/src/build/libc/sunrpc/xbootparam_prot.T /usr/local/src/build/libc/sunrpc/xbootparam_prot.c touch /usr/local/src/build/libc/sunrpc/xbootparam_prot.stmp ) I inquired on the gcc mailing list to determine if this is/was a problem... Here's the start of the thread: http://gcc.gnu.org/ml/gcc/2001-06/msg00063.html Here's the most recent reply: http://gcc.gnu.org/ml/gcc/2001-06/msg00075.html Basically, "glibc will have to be updated for GCC 3.0" So testing released builds of glibc is not very worthwhile for checking the integrity of gcc-3.0 until glibc is updated -(. -- Gordon Sadler
