Skip to content

LEARNING_PATH.md

Latest commit

 

History

History
141 lines (93 loc) · 4.57 KB

LEARNING_PATH.md

File metadata and controls

141 lines (93 loc) · 4.57 KB
Raw

The 7-day learning path

One focused hour a day. Read, run, then do the exercise. Do not skip the exercises — reading automation code feels productive and teaches almost nothing.

Day 1 — Setup and orientation

Read: 00-setup.md

make install
make smoke          # should be green in ~30 seconds
make app            # then open http://127.0.0.1:5000 and click around

Poke the app manually first. You cannot automate what you have not used.

Checkpoint: you can explain what make smoke selected and why it was faster than make test.

Day 2 — Playwright fundamentals

Read: 01-playwright-basics.md Code: tests/01_basics/ — five files, read them in order.

pytest tests/01_basics -v
pytest tests/01_basics/test_02_locators.py --headed --slowmo 700

Exercises 1–5 are at the bottom of each lesson file.

Checkpoint: you can say, without looking, why expect(loc).to_have_text("x") is safer than assert loc.inner_text() == "x".

Day 3 — Page Object Model

Read: 02-page-object-model.md Code: pages/, then tests/02_pom/

pytest tests/02_pom -v

Exercise 6: the app has an upload page with a page object already (pages/upload_page.py) but no POM tests. Write tests/02_pom/test_upload.py covering: upload from disk, upload from memory, and the error when you submit with no file.

Checkpoint: a designer renames the "Log in" button to "Sign in". How many files do you change?

Day 4 — Fixtures, markers, and data

Read: 03-fixtures-and-markers.md

pytest --fixtures | grep -A3 "authenticated_page\|api_context\|storage_state"
pytest --markers
pytest -m "smoke and not slow" -v

Exercise 7: add an admin_page fixture that is logged in as admin rather than demo, with its own storage state, and write one test that uses it.

Exercise 8: add a perf marker to pytest.ini, tag one test with it, and make it run in neither the smoke nor the regression selection.

Checkpoint: you can explain why configure_timeouts uses request.fixturenames instead of just taking page as an argument.

Day 5 — CI/CD (the big one)

Read: 04-ci-cd-explained.md with .github/workflows/ci.yml open beside it.

Exercise 9: in your fork, create a branch, deliberately break one assertion, and open a PR. Watch the pipeline. Then:

  1. Find which shard the failure landed in.
  2. Download that shard's artifact and open report.html.
  3. Open the trace: playwright show-trace trace.zip.
  4. Read the bot's PR comment and the inline failure annotation.
  5. Fix it and watch the comment update rather than duplicate.

Exercise 10: change the matrix so WebKit runs only on main, not on pull requests. (Hint: include plus an if: on the step, or a second job.)

Checkpoint: you can explain what concurrency: cancel-in-progress saves and what it risks.

Day 6 — Debugging

Read: 05-debugging-traces.md

make debug                          # Playwright Inspector
playwright codegen http://127.0.0.1:5000    # record a script, then improve its locators
make trace

Exercise 11: run playwright codegen, record a login + filter flow, then rewrite the generated code using page objects and role-based locators. Compare the two. Codegen is a starting point, not an output.

Checkpoint: given only a trace.zip from a CI failure, you can say what the page looked like at the moment of failure.

Day 7 — Flakiness and strategy

Read: 06-flaky-tests.md

pytest -m regression -n auto        # run it three times
pytest -m regression -p no:randomly # and once in file order

Exercise 12: introduce a flaky test on purpose — make two tests share a hard-coded user ID — then run with -n 4 until it fails. Fix it properly (unique data), not with --reruns.

Checkpoint: you can argue for what belongs in smoke versus regression on a real product, and defend the size of your smoke suite.

Where to go next

  • Swap the bundled app for your own product; keep pages/, utils/, conftest.py and the workflows
  • Add contract tests against your real API and wire them into the api-tests job
  • Add accessibility checks (axe-core via page.evaluate) as a new marker and a new CI job
  • Add a staging environment in GitHub with a required reviewer, and point manual-run.yml at it
  • Replace pytest-html with Allure if your team wants historical trends