http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47196
Tobias Burnus <burnus at gcc dot gnu.org> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|UNCONFIRMED |RESOLVED Component|bootstrap |libfortran Resolution| |WONTFIX --- Comment #1 from Tobias Burnus <burnus at gcc dot gnu.org> 2011-01-07 08:52:01 UTC --- I would suggest that you read the libquadmath related part of http://gcc.gnu.org/install/configure.html If you do not want to use libquadmath in libgfortran, build with --disable-libquadmath-support -- or supply a system version of libquadmath library/header files. Whether libquadmath is build (as part of GCC) is independent from whether Fortran (gfortran, libgfortran) uses libquadmath or not. This allows for the following four combination: a) Build libquadmath and Fortran with libquadmath-based REAL(16) support (default on systems where __float128 is supported) b) Build neither libquadmath nor libquadmath-based REAL(16) support (obviously the default on systems lacking __float128 support; Other use: Save a few kilobytes by having one library less) c) Build libquadmath but no Fortran REAL(16) support If one wants to save a few KiB for Fortran but wants to us libquadmath, e.g, with C/C++ d) Don't build libquadmath, but support libquadmath-based REAL(16) in gfortran Useful only if the system has already a (compatible) libquadmath installation libgfortran then links "-lquadmath" but the user/builder has to make sure that it is in the -I and the -L paths.