Software Quality and Quality Assurance

Why Smarter. Quality is usually taught in fragments: a unit testing chapter here, a linter there, a CI tutorial on a project with three tests. Students rarely see a complete quality process that already works on a large codebase, or learn what each layer catches and what slips through. Smarter has one, and every layer is in the repository:

  • Tests at scale. More than seven thousand unit tests across Python and React, built on time-test code infrastructure like SmarterTestBase and shared mixins that test every resource’s broker, API and views the same way.

  • Test doubles that protect production. The development containers hold live cloud credentials, so an in-memory provider replaces every DNS, certificate, Kubernetes and email call in tests, and a custom test runner skips the tests that need real infrastructure unless they are asked for.

  • Coverage with a target. At least 90% for each Django app, 95% where that takes moderate effort, measured with make coverage and reported to Codecov under separate Python and React flags.

  • Checks before every commit. The pre-commit hooks, black, isort, flake8, autoflake, pyupgrade, pydocstringformatter, bandit, codespell, Prettier and ESLint, secret detection and commitlint, are the project’s one lint standard (ADR 028).

  • CI and releases. A test workflow that rebuilds the developer environment step by step, with caches and no cloud credentials, separate workflows for builds, chart releases and deployments, and semantic-release that turns conventional commits into versions and changelogs.

  • Quality when agents write the code. Most new code is written by coding agents, against a definition of done in CLAUDE.md and the testing skills, so students can study how a quality process holds up when the author is a model.

Students see a complete quality process that already works, rather than assembling one from fragments, and can measure where it is strong and where it is not.

What the course covers. What quality means: defects and their cost, the test pyramid and shifting left; writing good unit tests; testing Django, with fixtures and cleanup without transaction rollback; test doubles, and where to patch; integration tests, Celery tasks and tagged tests; front end tests with Vitest, Testing Library, Storybook and MSW; static analysis and the pre-commit hooks; code review, including code written by AI agents; documentation as quality; CI with GitHub Actions, caching and matrix builds; conventional commits and automated releases; and measuring quality with coverage, mutation testing, flaky test detection and the DORA metrics.

Look at.

Each course plan lists every file that it uses under Smarter in the Source Code.

Read. Unit Testing, CI/CD, Semantic Versioning Guide, ADR 013 on testing, ADR 028 on code quality.

Exercises.

  • Run make coverage for one Django app, find its least covered module, and raise its coverage with meaningful tests.

  • Introduce a deliberate bug, and see which tests, hooks and CI jobs catch it, and which do not.

  • Fork the repository, add a push or pull_request trigger to test.yml, which upstream starts by hand, and get it passing on the fork. Use it as an autograder for later assignments.

  • Review a pull request written by a coding agent, and decide whether each of its tests checks behavior or only executes lines.

Course plans. Three complete, ready-to-teach plans, each with a schedule, graded assignments and rubrics, two capstone projects, recommended texts, further study and technical prerequisites. See Course Formats and Grading for what the formats have in common.

See the Smarter Instructor Guide to install Smarter and set up a class.