Cover the patient with tests
Part 2 of 4 learning hours on this topic. All are specific for the C language.
- define a Stub
- recognize link-time substitution
- use stubs in unit tests
- 5 min connect: When do you have enough tests?
- 10 min concept: stubs and link-time substitution
- 35 min concrete: cover the patient
- 5 min conclusions: confidence in these tests?
Connect - when do you have enough tests?
How do you know you have enough unit tests? Put a mark next to any you agree with.
- When the Architect is happy
- When coverage is 80%
- When I get fed up of writing tests
- When I feel confident the code works
- When I’ve written the tests on my test list
- When Jupiter is in Ares and Saturn is in Scorpio
- When I no longer have the feeling there are bugs in the code I’m looking at
- When I can run the code and it seems to work
- When I’m confident the tests will survive a refactoring of the code
Concept: Break dependencies using stubs
Last time we identified one problem with adding tests to existing code is to break dependencies. In the exercise repo Grep with Marketing there is a branch ‘with_stubs_no_tests’. It has the patient in the surgery, and each of the three dependencies is replaced by a stub implementation. Review the stub code.
Explain what a stub is and what problem it solves. Show how the link-time substitution works.
Concrete: add tests
Write tests for the patient, using the stubs. Aim for 100% code coverage.
In your own words, write down what a stub is and when you might use one.