Session 07 — Regression Check, one step at a time

Back to the lab · All help guides

Finish with: a small test diff, actual red/green results, locator rationale and coverage limit. Use 35 minutes. If browser automation is blocked, use the lab's assertion-design route and label execution unverified.

1. Open the starter and requirement

Read R3 in the requirements. In your local course copy, open sandbox/starter.spec.mjs. It checks a heading after clicking Load allocation. Ask yourself what that fails to establish about the record.

Create local submissions/session-07 for your brief, notes and run results. Only the learner starter test should change; preserve the deliberately buggy candidate, fixed comparison and requirements.

2. Prepare the two terminals

In terminal 1, run npm start and keep it running. In terminal 2, follow the optional Playwright setup. On Windows PowerShell use npm.cmd and npx.cmd if needed. Linux browser dependencies may need instructor help; platform guidance explains the supported route.

Checkpoint: http://127.0.0.1:4178 opens and the starter can run. A green heading-only starter is not sufficient evidence of a successful allocation journey.

3. Plan the assertion before editing

Copy the task brief. Tell Codex to inspect the starter, R3 and relevant local markup/API code, explain the weakness, then propose the smallest patch. Ask for a stable selector, an observable reserved-record assertion and a relevant response/status condition. Review the proposal yourself before applying it.

Do not make the expected outcome weaker to obtain a pass. A wait for the relevant response must be registered before the click that triggers it; fixed sleeps and repeated retries can hide defects.

4. Run the same changed test against both candidates

Use terminal 2. Mac/Linux:

CANDIDATE=buggy npx playwright test sandbox/starter.spec.mjs
CANDIDATE=fixed npx playwright test sandbox/starter.spec.mjs

Windows PowerShell:

$env:CANDIDATE = "buggy"
npx.cmd playwright test sandbox/starter.spec.mjs
$env:CANDIDATE = "fixed"
npx.cmd playwright test sandbox/starter.spec.mjs
Remove-Item Env:CANDIDATE

Windows Command Prompt:

set "CANDIDATE=buggy"
npx.cmd playwright test sandbox/starter.spec.mjs
set "CANDIDATE=fixed"
npx.cmd playwright test sandbox/starter.spec.mjs
set "CANDIDATE="

Run lines separately and save the results before continuing. These environment settings select runtime candidates; they are not Git branches. If using an alternate port, set ACADEMY_PORT in both terminals using the platform instructions.

Checkpoint: the improved check fails A for the intended allocation problem and passes B. Installation, timeout or connection errors do not count as a meaningful red result. npm test does not run this Playwright test.

5. Save and check your work

Review the before/after test diff, stable-locator explanation and one uncovered behavior. Git users can inspect git diff -- sandbox/starter.spec.mjs; ZIP users can compare a preserved starter copy in their editor. Include the diff and genuine run evidence in the practice packet.

If you get stuck

If installation cannot be completed, write the exact selector, expected record, response/status condition, expected A/B behavior and why it detects the defect. Mark runs unverified and arrange hands-on follow-up. Never invent logs. Help checklist.