Building libcmaes on Mac OSX

Building libcmaes

Instructions for compiling on OSX Maverick (10.9) [untested on other versions] with Mac Ports. Similar instructions should work for Homebrew users as well:

• install macports and export PATH:

export PATH=$PATH:/opt/local/bin:/opt/local/sbin:/usr/bin:/usr/local/git/bin

• install gcc-47 or higher (gcc-48, gcc-49) from macports

sudo port install gcc48
sudo port install autoconf
sudo port install automake

• install eigen3 from macports

sudo port install eigen3

• install gflags from macports:

sudo port install gflags +gcc48

• get libcmaes from github:

git clone https://github.com/beniz/libcmaes.git

• configure libcmaes:

./autogen.sh
./configure CC=/opt/local/bin/gcc CXX=/opt/local/bin/g++ --with-eigen3-include=/opt/local/include/eigen3 CXXFLAGS="-lstdc++ -I/opt/local/include -O4" LDADD=-lgflags LDFLAGS=-L/opt/local/lib

• build libcmaes

make

• test libcmaes

./tests/test_functions

should return results of a run on fsphere in 2-D.

Building the library with Python bindings

Python bindings are yet not part of the production code, so switch to the dedicated branch:

git checkout python

The bindings have dependencies on:

• boost-python
• numpy Install both as needed:

sudo port install boost +gcc48
sudo port install py-numy

Configure and build:

./autogen.sh
./configure CC=/opt/local/bin/gcc CXX=/opt/local/bin/g++ --with-eigen3-include=/opt/local/include/eigen3 CXXFLAGS="-lstdc++ -I/opt/local/include -O4 -I/opt/local/include/" LDADD=-lgflags LDFLAGS="-L/opt/local/lib" BOOST_LDFLAGS=-L/opt/local/lib --enable-python
make