Jonathan Wakely wrote: > > On Sun, Jan 13, 2002 at 02:50:06PM -0800, Stan Shebs wrote: > > > Making all in libmath > > /bin/sh ../libtool --mode=compile /Users/shebs/i/gcc/base/x~1/gcc/xgcc >-B/Users/shebs/i/gcc/base/x~1/gcc/ -B/usr/local/powerpc-apple-darwin5.1/bin/ >-B/usr/local/powerpc-apple-darwin5.1/lib/ -isystem >/usr/local/powerpc-apple-darwin5.1/include -DHAVE_CONFIG_H -I. >-I../../../../gcc/libstdc++-v3/libmath -I.. -g -c >../../../../gcc/libstdc++-v3/libmath/signbit.c > > libtool: compile: unable to infer tagged configuration > > libtool: compile: specify a tag with `--tag' > > make[3]: *** [signbit.lo] Error 1 > > I see exactly the same problem on i586-pc-linux-gnu, if I use "--prefix=~" > to configure the build. > > I've never seen it said that --prefix=~ isn't allowed (although I can see > that if the arg is never interpreted by the shell it won't get expanded) so > is it supposed to work? If not, is there a way to prevent me from using it, > or to make it clear why the build has failed?
The problem is that libtool actually tries to do the right thing for '~' and other chars in paths, by looking for known metachars and enclosing such arguments in "", but the tag inference machinery is comparing a $base_compile with quoted arguments to a value of CC pasted in by ltconfig that does not have anything quoted. So the string search fails. I've been able to get the build to finish by tweaking the generated libtool's CC directly, but it's not yet clear whether it's better to fix this by quoting CC sometime before it gets compared to base_compile, or to strip the double quotes from base_compile when comparing to CC. My guess is the former, but I'd be glad of guidance from experienced libtoolers. Stan _______________________________________________ Libtool mailing list [EMAIL PROTECTED] http://mail.gnu.org/mailman/listinfo/libtool
