This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: Patch installed for warning regression fp-bit.h prototype
> From: Alexandre Oliva <aoliva at redhat dot com>
>
> On Mar 3, 2003, "Kaveh R. Ghazi" <ghazi at caip dot rutgers dot edu> wrote:
>
> > Was it supposed to fix something that's currently demonstrably
> > broken, or something that's only broken when combined with my
> > attempts to use the irix6 native long double libc functions?
>
> The latter. Actually, it also fixes the former if you happen to call
> libc functions that return long doubles that might violate the
> assumption that both doubles in the pair have the same sign, which is
> what this patch fixes.
>
> > I'm testing that patch by itself against a clean 3.3. If it's meant
> > to be used in conjunction with the native float libs patch, then
> > perhaps my particular test isn't worthwhile. Let me know.
>
> As long as it doesn't introduce any regressions, it will be a strict
> improvement, even if for conditions that our testcase can't exercise.
Ok the test is complete. Patch applied to clean 3.3 branch sources,
bootstrapped with no regressions on mips-sgi-irix6.5.
I'll dig out my work on the native float libs, but that's take a
while just to recapture my working context. In the mean time, I
suggest installing the patch on trunk and 3.3.
Thanks,
--Kaveh
--
Kaveh R. Ghazi ghazi at caip dot rutgers dot edu