AMDG On 11/27/2011 04:50 AM, bdawes@acm.org wrote:

Author: bemandawes Date: 2011-11-27 07:50:00 EST (Sun, 27 Nov 2011) New Revision: 75690 URL: http://svn.boost.org/trac/boost/changeset/75690

Log: Change deprecated /Op option to its replacement, /fp:precise. This quiets a warning from the Intel compiler, which uses the msvc toolset.

Did you miss my reply on boost-build? Please revert this. It's wrong.

On 11/26/2011 06:03 AM, Beman Dawes wrote:

...

...

The patch below clears a warning from Intel 12.1, and works fine for MSVC 8, 9, and 10.

Unless someone objects, I'll apply it tomorrow. I'm posting advance notice in case any of the math folks want to give it a try beforehand.

I object. This branch is for msvc 6 and 7, which only have /Op. /fp:precise is for msvc 8 and up. The real problem is that intel-win is assuming vc7.1 compatibility.

Text files modified: trunk/tools/build/v2/tools/msvc.jam | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-)

Modified: trunk/tools/build/v2/tools/msvc.jam ============================================================================== --- trunk/tools/build/v2/tools/msvc.jam (original) +++ trunk/tools/build/v2/tools/msvc.jam 2011-11-27 07:50:00 EST (Sun, 27 Nov 2011) @@ -240,7 +240,7 @@

# Improve floating-point accuracy. Otherwise, some of C++ Boost's "math" # tests will fail. - toolset.flags $(toolset).compile CFLAGS $(conditions) : /Op ; + toolset.flags $(toolset).compile CFLAGS $(conditions) : /fp:precise ;

# 7.1 and below have single-threaded static RTL. toolset.flags $(toolset).compile CFLAGS $(conditions)/<runtime-debugging>off/<runtime-link>static/<threading>single : /ML ; _______________________________________________ Boost-commit mailing list Boost-commit@lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-commit