http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56887
--- Comment #3 from fkrogh at mathalacarte dot com 2013-04-09 13:30:45 UTC --- On 04/09/2013 06:06 AM, burnus at gcc dot gnu.org wrote: > http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56887 > > Tobias Burnus <burnus at gcc dot gnu.org> changed: > > What |Removed |Added > ---------------------------------------------------------------------------- > CC| |burnus at gcc dot gnu.org > > --- Comment #2 from Tobias Burnus <burnus at gcc dot gnu.org> 2013-04-09 > 13:06:15 UTC --- > (In reply to comment #0) >> tests for equality between reals is flagged with a warning. > The support for the warning follows ISO/IEC TR 24772, which recommends users > to > "Avoid creating a logical value from a test for equality or inequality > between two floating-point expressions." - And compiler vendors to provide > such > a warning. > > The problem with many constructs is that one cannot reliably detect whether > they are okay or a bug in the code. Thus, warnings are issued for those; there > are always false positives and missed bugs with warning diagnostic. Those > checks, where the false-positive rate is low and the likelihood for bugs is > high, are enabled by default; others only with -Wall or -Wextra or only with > -W<name-of-the-warning>. > > In your case, using integer-valued floating-point numbers is probably fine. As > Thomas wrote, you can use -Wno-compare-reals to disable the warning. See GCC > 4.8 release notes or gfortran's man page (or user manual) under -Wextra. > Many thanks Tobias. That works fine for me and is just what I wanted. For some reason it is not in my info pages or man pages -- I really did look for anything like this. On another subject, I know that you had some contact with Dr. Richard Hanson concerning IEEE exceptions. He has completed his work on these and is free to tun them over to GNU. He is under the impression that there is no interest on your part. Thanks again, Fred
