Skip to content

exercises.md

Latest commit

 

History

History
94 lines (61 loc) · 3.94 KB

exercises.md

File metadata and controls

94 lines (61 loc) · 3.94 KB

Exercises

Prelim / setup

  1. Clone the workshop repo and navigate to the example directory
git clone https://github.com/Cambridge-ICCS/testing-workshop
cd testing-workshop/exmaple
  1. Setup a virtual environment (optional step) and install the dependencies:
python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt
  1. (Optional) To make it easy to import the Python module for exploration, set the PYTHONPATH to include the src directory:
export PYTHONPATH=$PYTHONPATH:$(pwd)/src
  1. Make sure you can run the smoke tests to check all is working well:
pytest tests/smoke_test.py

Unit tests

  1. Run the unit tests in test_energy_balance.py
pytest tests/test_energy_balance.py

The temperature_at_energy_balance takes two (float) arguments, one for albedo of the Earth (i.e., a value between 0 and 1) and one for the emissivity (ratio of radiation actually emitted versus energy emitted if it were a true black body).

  1. Write a unit test for temperature_at_energy_balance that makes two input-output assertions based on some extremal points for which it is possible to work out what the behaviour without running the code.

  2. Add a further test of temperature_at_energy_balance input-output behaviour by picking some other input values, running the code (i.e., in the repl) to see the output, and then codifying this in a test.

    Don't forget to run the init() function first when experimenting, but use the setup text fixture to have this done for you in the tests.

  3. Turn the above three input-output tests into a 'parameterized' test (see test_helpers_param.py for an example) to avoid repetition.

  4. (Bonus) Check the coverage of the whole test suite:

pytest tests/ --cov src

You can get a report on which lines are not visited by any tests with:

pytest tests/ --cov src --cov-report term-missing

Or if you prefer more visual report, you can try to generate HTML report

pytest tests/ --cov src --cov-report html

this will generate htmlcov folder where you can see the report in your browser by opening htmlcov/index.html

Choose a section of lines not covered by a test and write an additional test to increase the coverage. Verify that this has improved coverage aftwerwards.

Property-based testing

  1. The energy_out function (from the energy_balance module) determines the energy radiated from the Earth as a function of its temperature (in Kelvin) assuming it is a black body emitter. What is a property that we can always expect of the output? Write a property-based test for this.

    Consider what strategy (generator) to use and how you can provide suitable inputs to the test.

The plot_temperature_vs_emissivity() and plot_temperature_vs_albedo() functions generate matplotlib plots and use a custom linspace function in helpers.py where linspace(x, y, n) generates a NumPy of n floating point values from x to y inclusive, evenly spaced. Now for the linspace function:

  1. Write some post-condition-style property-based tests that relate values in the output to the input arguments. Hint: think about whether arguments should appear in the output, and where, and how the size of the output relates to the arguments. Hint 2: remember that you can limit the range of generated floats with max_value and min_value keyword arguments to the generator.

  2. Think of what a valid output should look like for linspace. Codify this as a property based test.

  3. Consider what happens when the first two arguments are flipped and from this create a further general property-based test for linspace. Hint: you may also need the np.flip function which reverse an array.

Extension: Doctests demo and exercise

  1. A 'doctest' is a kind of comment that contains an example that is then machine-checked as a test. Navigate to example/src/doctest_demo.py to see an example, with explanation, and an in situ exercise.