Skip to content

Latest commit

 

History

History
57 lines (31 loc) · 1.93 KB

app-course-pr-scenarios.md

File metadata and controls

57 lines (31 loc) · 1.93 KB

Book App Web PR Scenarios

Use these drafts to create repeatable pull request exercises for the course.

PR scenario 1: Review comment asks for clearer empty-state copy

Branch: fix-empty-state-copy

PR summary

This PR updates the empty state shown when filters match no books.

Seeded PR conversation comment

The copy is better, but can we make it more helpful for a first-time learner? Please mention that they can change the search term, genre, or reading status.

Learner goal

Ask Copilot to address the PR conversation comment, inspect the diff, and decide whether to accept the change.

PR scenario 2: Failing CI points to the stats test

Branch: practice-failing-stats-check

PR summary

This PR intentionally changes the stats logic so CI can demonstrate a failing check.

Seeded CI failure

npm test -- --run fails in src/tests/stats.test.tsx because the favorite count excludes unread favorites.

The repository includes .github/workflows/book-app-web.yml, which installs dependencies with npm ci, runs the tests, and builds samples/book-app-web. Use this workflow for the failing-check lesson after creating the intentional regression branch.

Learner goal

Use the failing check as context, ask Copilot for a root cause, and fix the stats logic without changing the test expectation.

PR scenario 3: Agent Merge waits on checks and review state

Branch: feature-reading-dashboard

PR summary

This PR adds a safe training note used for merge-readiness discussion after tests pass and comments are resolved.

Merge readiness checklist

  • Tests pass with npm test -- --run.
  • The app builds with npm run build.
  • The Book app web workflow passes in GitHub Actions.
  • Review comments are resolved.
  • The learner confirms the branch is safe for a training repository.

Learner goal

Discuss when Agent Merge is appropriate, what it waits for, and why a human should still review the final diff.