------- Comment #46 from rguenther at suse dot de 2009-08-26 09:34 ------- Subject: Re: Problems with -fwhole-file
On Wed, 26 Aug 2009, jv244 at cam dot ac dot uk wrote: > ------- Comment #45 from jv244 at cam dot ac dot uk 2009-08-26 09:23 ------- > Paul, are there issues other than the ones reported in comment #44 that > prevent > making -fwhole-file the default? > Richard, do you believe PR38913 (Fortran does not set TYPE_CANONICAL properly) > can be fixed for 4.5? I don't really know. > I believe whole-file should be made default before the end of stage 1, if any > serious problem emerges that can not be fixed, one can always do a step back. True. > For example, LTO (-flto) enables -fwhole-file by default, and presumably will > be merged in trunk one of these days. > > Furthermore, I believe that it would be a good idea to enable -fwhole-file by > default for Fortan anyway, so that the additional, useful, warning messages > are > generated. This will furthermore guarantee that the number of warnings > generated by the FE doesn't depend on optimization flags such as -flto. A > bonus > is of course that whole-file gets more testing, so that it becomes reliable. Correct. If required I can build SPEC2000 and SPEC2006 with -fwhole-file again and see what the fallout is. Richard. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40011
