The project employs a diverse suite tests that help ensure it works as intended and prevents regressions as it continues to grow and evolve.
Most of the internal components the project uses have unit tests to thoroughly test them. Here dependencies of components are mocked or stubbed appropriately to ensure tests are reliable, test only one component and are fast!
Acceptance tests run the built shopify command line against a wide range of fixtures and verify its output and results. They are the slowest to run however provide the most coverage. The idea is to test a few complete scenarios for each major feature.
Those are written in Cucumber and Ruby and can be found in the features/ directory. They can be executed by running:
bundle exec cucumber
bundle exec cucumber features/theme.feature:3 # A specific test
Note that we currently don’t have an approach for stubbing the interactions with the GraphQL APIs and that therefore we can’t write acceptance tests for commands that interact with APIs.
When developing acceptance tests, it can be helpful to see the outputs of running commands.
To see outputs, append --verbose when running an acceptance test. Example:
bundle exec cucumber features/theme.feature --verbose