Contributing

Development happens on github and the preferred contribution method is by forking the repository and issuing a pull request. Alternatively, just sending a patch to luis@luispedro.org will work just as well.

If you don’t know git (or another distributed version control system), which is a fantastic tool in general, there are several good git and github tutorials. You can start with their official documentation.

If you want to work on the C++ code, you can read the chapter in the internals before you start. Also, read the principles declaration.

Debug Mode

If you compile mahotas in debug mode, then it will run slower but perform a lot of runtime checks.

There are two levels:

1. A release build with b_ndebug=false turns on assertions. The code will run slower, but probably not noticeably slower, except for very large images.

2. make debug uses an optimized release build with assertions enabled and additionally defines _GLIBCXX_DEBUG. This only has an effect with libstdc++, but when available it enables checked iterators in the C++ standard library. The result can still be much slower as many iterator operations are now checked. However, it catches many errors.

The Makefile that comes with the source helps you:

make clean
make debug
make tests

will rebuild in debug mode and run all tests. When you are done testing, use the fast Make target to get the non-debug build:

make clean
make fast

Using make will rebuild the editable install in your current Python environment with the requested configuration.

If you don’t know about it, check out ccache which is a great tool if you are developing in compiled languages (this is not specific to mahotas or even Python). It will allow you to quickly perform make clean; make debug and make clean; make fast so you never get your builds mixed up.