The tests directory contains simple unit testing scripts that were used during development, and should still run after code has been added or changed.
We wish to use pytest.
Currently, unit testing is largely incomplete, but it will probably detect new changes to the code that break existing functionality. For this type of scientific code, functional testing (i.e. using 'real' benchmark test cases and verifying the overall result) is much more important than unit testing.
You should run pytest from the project root directory. This runs all tests. You can run pytest on specific tests by invoking pytest tests/test_<name-of-the-test>.py
Before doing pytest, you may also directly run the tests/test_<name-of-the-test>.py script in Python using python -m tests.test_<name-of-the-test> (without the .py extension) so that you can inspect the console output before applying the pytest.
All tests are best run from the project's root directory (i.e. have this root directory as the current working directory). Running the script from within the tests directory may not work (problems with importing modules). Some tests, however, can be run with tests as the working directory so that they can be edited and run using Spyder (these have a work-around for the import problem).
- The test scripts are presently really verbose, displaying graphs and generating ample stdout output. There are also a number of redundant parts in the scripts. The test scripts need more actual
test_somethingfunctions withassertinstead of just running the code and plotting graphs. - Create 'silent' testing! (Perhaps as an option to be supplied, or just in a general sense)