--- Additional Comments From ebotcazou at gcc dot gnu dot org 2005-02-12
19:20 ---
It's a bug in intrinsics/c99_functions.c:nextafterf. Fixing.
--
What|Removed |Added
--- Additional Comments From coudert at clipper dot ens dot fr 2005-01-24
07:30 ---
I confirm comment #3: PR 17675 is fixed, but this bug still appears on the
nightly test logs I do on sparc-sun-solaris2.9. Bug or not, this should be fixed
(either the compiler or the testcase).
I inclin
--- Additional Comments From coudert at clipper dot ens dot fr 2005-01-07
12:30 ---
(reply to comment #4)
After your comment, not sure either if it's a bug. The original testcase
(intrinsic_nearest.f90) fails with Intel (7 and 8) and Portland Group compilers
on i686, but works with g95
--- Additional Comments From Thomas dot Koenig at online dot de 2005-01-07
10:56 ---
I am not sure that this is indeed a bug.
Is the result of nearest for +/- infinity defined to
be +/huge ?
Look at what ifort does:
$ cat near-inf.f90
program test_nearest
real inf
inf = exp(1000.)
--- Additional Comments From ebotcazou at gcc dot gnu dot org 2005-01-07
00:33 ---
> I think this is a dup of bug 17675.
I'm not sure: I've Steve's patch for PR 17675 in my tree and I still see the
failures everywhere.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19302
--- Additional Comments From ebotcazou at gcc dot gnu dot org 2005-01-07
00:32 ---
Confirmed on all SPARC/Solaris platforms (32-bit or 64-bit).
--
What|Removed |Added