Python封装的性能研究

Python Wrapper Tools: A Performance Study

● Manual wrapping is still fastest and most versatile

● SIP, Boost and Babel generate native extension modules with low overhead

● SWIG-generated modules rely on pure Python wrapper module and introduce a large overhead

● Wrapper tool highlights

○ SIP generates the fastest wrappers

○ SWIG is mature and well documented

○ Boost.Python most elegant integration of C++ and Python

○ Babel supports languages as both target and source