This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [gfortran] Let f951 generate type infos for SELECTED_*_KIND
- From: Paul Brook <paul at codesourcery dot com>
- To: fortran at gcc dot gnu dot org
- Cc: Tobias Schlüter <tobias dot schlueter at physik dot uni-muenchen dot de>,patch <gcc-patches at gcc dot gnu dot org>
- Date: Tue, 5 Oct 2004 00:07:10 +0100
- Subject: Re: [gfortran] Let f951 generate type infos for SELECTED_*_KIND
- Organization: CodeSourcery
- References: <415FD810.3000502@physik.uni-muenchen.de> <4161D4B4.4070705@physik.uni-muenchen.de>
On Monday 04 October 2004 23:54, Tobias Schlüter wrote:
> Tobias Schlüter wrote:
> > This straightforward patch implements this. The only non-trivial part was
> > finding out how to pass the new option to f951 during the build, but this
> > turned out to be trivial as well, after I had found the correct driver
> > option.
>
> Unfortunately, I hadn't thought aobut multilibbed targets. These call the
> same compiler, but with different options, e.g. 'f951 -m32' vs. 'f951
> -m64'. This of course breaks my patch. There's no -Wc options, which would
> allow running "$(F77COMPILE) -Wc,-output-real-type-infos", and I also
> didn't find another solution. If anybody knows an elegant solution for how
> to run the compiler as opposed to the compiler driver with the correct
> options, I'm listening.
Can't you rename the option to -fsomething? That should be passed properly, so
you won't need the --print-prog-name= stuff. You can just do gfortran
-fsomething -o file.inc.
Maybe look how the -M options work in the C frontend.
Paul