I've read several threads about some of the differences between Cython and Swig and have implemented both techniques, but I am still not sure what route is the best route to take. Basically I have been working on a C++ library and I would like to expose those functions to Python. Cython seemed rather complicated and required me to write several .pyx files for every C++ class that I needed to expose whereas SWIG just did the work for me. I don't have a lot of experience with either one of these wrapping methods to date, but it seems that SWIG is a clear winner for wrapping my C++ code. So what is the controversy? Why would I spend hours writting .pyx files in Cython to wrap my C++ when I could just write one simple .i file in SWIG? Furthermore, I use CMake as my build tool and it was WAY easier to build SWIG with CMAKE than with Cython. I feel that I must be missing something because even big projects like OpenCV are not using SWIG and before I dedicate my project to SWIG, I want to find out why I would rather use Cython or nothing at all. Here is a summary:
My project: C++ source code is primary
- Python is a nice to have, just want a path of least resistance to expose the majority of my C++ code.
- Writting wrapper code sucks because now I have two places to maintain an API--I just want one, the C++ code.
Any advice is greatly appreciated. Also, I was curious about xdress.org--seems like that project has died though.
