This package contains cpp code, so it should use only CXXFLAGS; from build.log: x86_64-pc-linux-gnu-g++ -o lib/mypaintlib_wrap.os -c -Wall -Wno-sign-compare -Wno-write-strings -march=native -pipe -O2 -Xlinker -export-dynamic -fno-strict-aliasing -march=native -O2 -g0 -fwrapv -march=native -pipe -O2 -g0 -fPIC -I/usr/include/glib-2.0 -I/usr/lib64/glib-2.0/include -I/usr/include/libpng14 -I/usr/lib64/python2.7/site-packages/numpy/core/include -I/usr/include/python2.7 lib/mypaintlib_wrap.cpp It uses first "-march=native -pipe -O2" and is ok, after it uses CFLAGS: "-march=native -pipe -O2 -g0" amd64box ~ # emerge --info | grep CFLAGS CFLAGS="-march=native -pipe -O2 -g0" amd64box ~ # emerge --info | grep CXXFLAGS CXXFLAGS="-march=native -pipe -O2"
The code contains C and CPP code. The makefile appends the C{XX}Flags independently. This is not easy to fix and it requires some build system re-design. So even this bug is valid I wont fix it because the the 98% code is in python and only two files are written in C and CPP. The overhead of wrong CFLAGS is minimal so it will make no difference. testing mypaint-0.9.1 # find . -name "*.c" ./mypaint_exe.c testing mypaint-0.9.1 # find . -name "*.cpp" ./lib/mypaintlib_wrap.cpp Thanks for reporting anyway :)
