boost-libs and CXXFLAGS (Re: devel/boost-bjam)

Mikhail T. mi+thun at aldan.algebra.com
Fri May 10 20:28:02 UTC 2013


On 02.05.2013 18:21, Baptiste Daroussin wrote:
> Thanks you your investigation and tests, I'm expecting for your commit then.
Though boost-libs are now built with the specified compiler, the compiler flags 
continue to be ignored. I don't necessarily mind the optimization level raised 
from -O2, that I specify, to -O3 that boost prefers, but I am also setting 
-march=native on my systems and this is getting dropped. For example:

    g++47 -ftemplate-depth-128 -O3 -finline-functions -Wno-inline -Wall -pthread
    -fPIC -DBOOST_ALL_NO_LIB=1 -DBOOST_CHRONO_DYN_LINK=1
    -DBOOST_LOCALE_DYN_LINK=1 -DBOOST_LOCALE_NO_POSIX_BACKEND=1
    -DBOOST_LOCALE_NO_WINAPI_BACKEND=1 -DBOOST_LOCALE_WITH_ICONV=1
    -DBOOST_LOCALE_WITH_ICU=1 -DBOOST_SYSTEM_DYN_LINK=1
    -DBOOST_SYSTEM_NO_DEPRECATED -DBOOST_THREAD_BUILD_DLL=1
    -DBOOST_THREAD_NO_LIB=1 -DBOOST_THREAD_POSIX
    -DBOOST_THREAD_THROW_IF_PRECONDITION_NOT_SATISFIED -DBOOST_THREAD_USE_DLL=1
    -DNDEBUG -I. -I/opt/include -c -o
    bin.v2/libs/locale/build/gcc-4.7.4/release/threading-multi/shared/message.o
    libs/locale/src/shared/message.cpp

Without the -march flag, the compiler is forced to assume, the generated code 
must work even on i386, and can not take advantage of the MMX and the SSE 
instructions available on modern CPUs...

Can anything be done about this? Thanks! Yours,

    -mi



More information about the freebsd-office mailing list