Testing

This library is a bit odd when it comes to testing since it’s much harder to isolate & differentiate between units/components. The reason boils down to the fact this library (written in C++) is working with python’s entry-points mechanism which requires multiple environments just for testing different edge cases.

Integration Tests

In order to test this library a small testing program was written in C++ with the help of the argparse library. This small utility exposes the C++ library as a command-line program which is then used by pytest for the testing itself. The integration tests utilize tox for an automated management of python virtual environments and simulates 2 environments scenarios:

An environment without entry-points registered to a specific group.
An environment with entry-points registered to a specific group.

Running the Integration Tests

Running the integration tests is quite easy and it can be done as follows:

# assume we're in the repo's top directory.
tox

And if you want a bit more verbosity then use:

# assume we're in the repo's top directory.
tox -v

One can also run the integration tests without tox directly on conda’s environment like this:

# assume we're in the repo's top directory.
cmake -S . -B build/integration-tests  -Wdev -Werror=dev -DCMAKE_INSTALL_PREFIX=$CONDA_PREFIX -DCMAKE_FIND_ROOT_PATH=$CONDA_PREFIX -DENABLE_TESTING=ON
cmake --build build/integration-tests --target install
pytest test/ --loader build/integration-tests/test/loader

Note

Be sure to remember & verify that pytest is installed into conda’s environment (when using tox that’s not an issue since tox installs it for us). This can be verified like this:

pip list | grep pytest

# or alternativly
pip show pytest

If pytest wasn’t installed then you can run:

pip install pytest