Tests

Idea

Unit tests are particularly important to developers to debug their code separately from other coupling components. These unit tests also offer an opportunity to quickly check whether the installation of the package was successful, as mentioned on the start page.

The remainder of this documentation page focuses on running these tests. The first section explains the procedure to run (almost) all tests at once, which is useful for the post-installation test. The second section contains information on how to include the unit test of a solver wrapper in the testing framework. Finally, it is explained how to run a single test file, test class or test method. The latter is useful in the development of a new solver wrappers to quickly test a single aspect of the wrapper without having to test all aspects together, which can be time-consuming.

Running unit tests

The unit tests in CoCoNuT uses the unittest module available in Python. To run the default collection of tests which evaluates in a matter of seconds, navigate to the coconut/tests/ directory and type following command in the terminal:

python3 run_tests.py

This command will run all tests excluding the solver wrapper tests that are not pure Python. These tests are excluded because they each take several minutes to complete. The above command is equivalent to

python3 run_tests.py -fast

It is equally possible to run all test by using the keyword -all as follows

python3 run_tests.py -all

Besides these two predefined keywords, it is also possible to use (a set of) arbitrary keywords. For example

python3 run_tests.py fluent abaqus

will run all tests that have the fluent or abaqus in its path: for example coconut.tests.solver_wrappers.fluent.test_v2023R1.test_move_nodes.

Starting from Anaconda 2023.09, to run a specific test, it is sufficient to provide the full path, for example

python3 run_tests.py solver_wrappers.fluent.test_v2023R1.TestSolverWrapperFluent2023R1Tube3D.test_partitioning

Here, only the test method test_partitioning is run in the TestSolverWrapperFluent2023R1Tube3D class. Similarly, to run all methods in a particular class, it is sufficient to include the path up to the class name (or simply the class name itself, if there exists no duplicate).

Besides the method of running tests explained above, it is also possible to use commands of the form

python3 -m unittest -bv

Here, the keyword discover is assumed silently and all unit tests in modules named test*.py will be looked for. It will recursively find all the test files with this pattern for all the folders containing an __init__.py_ file. The -b keyword suppresses any output generated by the part of the code that is being tested, unless an error occurs or a test fails. Finally, the -v keyword enables a higher verbosity, as such the user can see what test methods have been run. Further documentation on the Python unit test framework can be found on the Python documentation website.

To run a specific test for earlier Python versions, it is required to use

python3 -m unittest -bv solver_wrappers.fluent.test_v2023R1.TestSolverWrapperFluent2023R1Tube3D.test_partitioning

What if solver software is not available

If a solver wrapper test is run, but the software is not available, the test will automatically be skipped and the non-availability wil be given as the reason. The availability of software is checked using the solver_modules.py file.