Unit Tests

Rydiqule comes bundled with a suite of unit tests that confirm basic functionality of the various components and checks for robustness to erroneous or unexpected arguments. We strive to follow the testing methodology and practices employed by numpy. We agree with the stipulation made there, that

“Long experience has shown that by far the best time to write the tests is before you write or change the code - this is test-driven development”

pytest

Rydiqule takes advantage of the pytest testing framework to run unit and integration tests of the code base. The full test suite is run from the project base directory with the command

pytest

This command will run all tests in the tests/ subdirectory as well as docstring examples. These tests cover a wide range of functionality as well as a number of representative integrations that demonstrate how the code can be used to generate end results.

Marks

The tests are marked based on the type of test that is being performed, and pytest can be told to only run certain tests. For example, this command will only run tests relating to the steady state solving functionality:

pytest -m steady_state

You can also exclude a specific group of tests. For example, this command will exclude tests marked as slow.

pytest -m "not slow"

Marks specifications can be combined using standard boolean keywords as well. The following will run all the time tests that are not slow.

pytest -m "time and not slow"

The available marks can be listed using pytest --markers. The markers we use are

Table 1 Markers

Marker	Description
slow	Marks a test as taking a long time to run
high_memory	Marks a test needing a lot of RAM
steady_state	Marks a test as using the steady-state solver
time	Marks a test as using the time solver
doppler	Marks a test that incorporates Doppler averaging.
experiments	Marks a test that represents a full experiment.
solution	Marks a test of the Solution object and methods.
util	Marks a test of the ancillary utilities.
structure	Marks a test of the definition of the atomic system.
exception	Marks a test of error handling.
backend	Marks a test that uses an optional backend
dev	Used to temporarily mark a single test that is being developed so it can run independently.

Docstring Rounding

Pytest also runs all examples in the docstrings using python’s built-n doctest module. This module runs the provided code and does simple string matching of the output to confirm results. Given rydiqule is a numerical computation package that deals heavily with floating point numbers, there will be small differences in results based on platform and even exact dependency libraries used. To prevent spurious errors, doctest has been configured to ignore digits of higher precision in the result than what is provided in the source. For example, a computed result of 9.333453 will successfully match against the docstring value of 9.333.

Coverage

If you install the pytest-cov plugin, you can check code coverage of the tests by modifying the command to read.

pytest --cov=rydiqule

Durations

If you want to see which tests take the longest to complete, you can use the --durations=n flag to give the n longest time tests:

pytest --durations=3

Settings the durations flag to 0 will cause pytest to report the time taken for all tests run.