On 06/20/2010 12:21 PM, Ralf Wildenhues wrote: > The Fortran test coverage in Autoconf is far from complete. It is not > easy: the relevant macros are intertwined, and Fortran compilers vary > wildly. > > This patch is a test balloon really. So far tested with GCC and one > Intel version on GNU/Linux, and Sun suite on Solaris. It would need > lots more testing, but I don't have access to many Fortran compilers. > I wouldn't be surprised to see a fair amount of failures with them, > consider yourself warned.
I'm okay with releasing 2.66 in order to widen the feedback path of more Fortran setups - we can deal with failures as they are reported (this change isn't adding regressions, just exposing any long-time issues). So yes, please apply. And thanks for the effort, since I am rather illiterate at the moment when it comes to Fortran. > Coverage for Fortran compiler macros. > > * tests/fortran.at (AC_OPENMP and Fortran 77) > (AC_OPENMP and Fortran): Simplify, using AT_CHECK_CONFIGURE. > (AC_F77_DUMMY_MAIN usage, AC_FC_DUMMY_MAIN usage) > (AC_F77_MAIN usage, AC_FC_MAIN usage, AC_F77_FUNC usage) > (AC_FC_FUNC usage, AC_FC_SRCEXT usage, AC_FC_FREEFORM): New > tests. > * doc/autoconf.texi (Fortran Compiler): Use GNU coding style > on C code snippets. Add markers for tested examples. > Suggest AC_FC_FREEFORM for source file extensions which the > compiler might not natively support but which are accepted > with help from AC_FC_SRCEXT. Suggest AC_CONFIG_HEADERS for > setups using one of the AC_*MAIN macros. -- Eric Blake [email protected] +1-801-349-2682 Libvirt virtualization library http://libvirt.org
signature.asc
Description: OpenPGP digital signature
