This is the mail archive of the
fortran@gcc.gnu.org
mailing list for the GNU Fortran project.
Re: -ffast-math and gfortran
On Mon, Jun 05, 2006 at 05:30:54PM +0200, Fran?ois-Xavier Coudert wrote:
> >I therefore
> >like Steve's proposal to rename -ffast-math to something that sounds
> >more dangerous, like -frisky-math.
>
> That should only be done at the GCC level, not only for gfortran. We
> don't want to have two options that do the same thing, with different
> names depending wether you're using the C or Fortran front-end.
>
This won't happen at the GCC level. Too much inertia.
OTOH, a language frontend can overrule any common option.
I'll suggest that -ffast-math emit a long-winded explanation
of the potential problems with -ffast-math, point to the
-frisky-math option, and exit. Of course, people will start
to -frisky-math without bothering to investigate why it is
risky.
Anyway, this will be my last post to any thread involving
-ffast-math. Hopefully, someone, who uses this option,
will actually document its assumptions and transformations.
--
Steve