test（测试）

Test

Testing Checklist

Feature Testing:
- [ ] Happy path works (main user flow)
- [ ] Edge cases handled (empty, long, invalid data)
- [ ] Error messages clear and helpful
- [ ] Works on mobile (iOS and Android)
- [ ] Works in Safari, Chrome, Firefox
- [ ] Loading states show during waits
- [ ] Forms validate input
- [ ] Can't break it with weird input
- [ ] Back button works correctly
- [ ] Page refresh doesn't lose data

See TEST-SCENARIOS.md for detailed scenarios.

When to Test

Test when:

• Feature just built by AI
• Before deploying to production
• After fixing a bug (verify fix works)
• Users report issues (reproduce first)

Don't test:

• While AI is still building
• Before feature is complete
• Every tiny change (batch test features)

Rule: Build → Test → Fix → Test again → Deploy

Structured QA Workflow

Don't test randomly. Follow this sequence every time.

Step 1: Orient — What Exists?

Before testing, know what you're testing. Walk through your app and list every page and feature.

Map Your App:
- [ ] List every page (home, dashboard, settings, etc.)
- [ ] List every form (signup, contact, create/edit)
- [ ] List every button that does something
- [ ] List every integration (payments, email, third-party)

Step 2: Explore — Visit Everything Systematically

Go through every page, every button, every form. Don't skip pages you "know" work.

Tell AI:

Go through every page of my app. Click every button. Fill every form.
Submit every action. Tell me what's broken, what looks wrong, and what
behaves unexpectedly. For each issue, tell me:
- What page it's on
- What you did
- What happened vs. what should happen
- How severe it is (blocks users vs. cosmetic)

Step 3: Score — Rate Your App's Health

Rate your app 1-10 on each dimension. Be honest — this is for you.

App Health Score:
                                      Score (1-10)
Does it load fast? (under 3 seconds)  ___
Do all buttons/links work?            ___
Do forms submit correctly?            ___
Do error messages make sense?         ___
Does it work on mobile?               ___
Is the core flow completable?         ___
                              Total:  ___/60

50-60: Ready to ship
40-49: Fix the gaps, then ship
30-39: Significant issues — fix before showing users
<30:   Major problems — keep building before testing

Step 4: Fix and Re-Verify

After each fix, re-test two things: the thing that was broken AND the things that were working. AI fixes often break something else.

After Every Fix:
- [ ] Original bug is resolved
- [ ] Core signup/login flow still works
- [ ] Core action still works
- [ ] No new visual issues on the page you changed

Step 5: The Fresh Eyes Test

Have someone who's never seen your app try to sign up and complete the core action. Watch without helping. What surprises you?

• Where do they hesitate?
• What do they click that you didn't expect?
• Where do they get confused but don't say anything?
• Can they describe what the app does after using it for 2 minutes?

This test reveals more than a week of solo testing. The things you "know" are obvious often aren't.

Manual Testing Checklist

1. Test happy path
   - Does the main flow work?
   - Can user complete the task?

2. Test edge cases
   - What if field is empty?
   - What if text is very long?
   - What if user clicks twice?

3. Test on mobile
   - Open on real phone
   - Test main actions
   - Check layout doesn't break

4. Test in different browsers
   - Chrome (most users)
   - Safari (iOS users)
   - Firefox (some users)

5. Document issues
   - Screenshot the problem
   - Write exact steps to reproduce
   - Give to AI to fix

Testing Edge Cases

Always test these:

Empty data:

• What shows when no items in list?
• What happens with empty form field?
• Is placeholder/empty state clear?

Long data:

• Very long name (200 characters)
• Very long text (10,000 characters)
• Does layout break?

Invalid data:

• Invalid email format
• Negative numbers where not allowed
• Special characters in text field
• SQL characters ('; DROP TABLE)

Boundary cases:

• Exactly at limit (100 char limit, enter 100)
• Just over limit (enter 101)
• Zero/empty values

Tell AI:

Test these edge cases:
- Empty username: show "Required"
- Username too long (>50 chars): show "Max 50 characters"
- Username with spaces: show "No spaces allowed"
- Special characters: show "Letters and numbers only"

See EDGE-CASES.md for comprehensive list.

Testing on Mobile

Minimum mobile tests:

Mobile Testing:
- [ ] Page loads and looks correct
- [ ] Can tap all buttons (44px minimum)
- [ ] Forms work (keyboard appears)
- [ ] No horizontal scroll
- [ ] Images load and fit screen
- [ ] Navigation works
- [ ] Can complete main user flow

Test on:

• iPhone (Safari) - Most common iOS
• Android phone (Chrome) - Most common Android

Don't need to test:

• Every phone model
• Tablets (unless primary use case)
• Landscape mode (unless important)

Cross-Browser Testing

Priority order:

1. Chrome (65% of users) - Test thoroughly
2. Safari (20% of users) - Test main flows
3. Firefox (5% of users) - Quick check
4. Edge (5% of users) - Usually works if Chrome works

Common browser issues:

• Date pickers look different
• Flexbox behaves differently
• Scrolling momentum feels different
• Animations may not work on old Safari

Quick test:

Open in each browser:
1. Load homepage
2. Sign up / Log in
3. Complete 1-2 main actions
4. Check nothing is broken

Testing Forms

Form validation testing:

Form Testing Checklist:
- [ ] Required fields show error if empty
- [ ] Email validation works (format check)
- [ ] Password requirements enforced
- [ ] Can't submit invalid form
- [ ] Error messages clear and specific
- [ ] Success message shows after submit
- [ ] Form disables during submit (no double-submit)
- [ ] Errors clear when user fixes them

Tell AI:

Test form validation:
- Required field left empty: "This field is required"
- Invalid email: "Enter a valid email address"
- Weak password: "Password must be 8+ characters with 1 number"
- All valid: Allow submit
- Show errors inline, not alert()

Testing Authentication

Auth flow testing:

Auth Testing:
- [ ] Can sign up with valid info
- [ ] Can't sign up with existing email
- [ ] Can log in with correct password
- [ ] Can't log in with wrong password
- [ ] Can reset password via email
- [ ] Session expires after timeout
- [ ] Logout works (can't access protected pages)
- [ ] Can't access protected routes without login
- [ ] Redirect to login when session expires

Testing Integrations

Third-party services:

Payment (Stripe):

- [ ] Test card (4242 4242 4242 4242) processes
- [ ] Declined card shows error
- [ ] Receipt email sent
- [ ] Subscription status updates
- [ ] Can cancel subscription

Email (SendGrid):

- [ ] Welcome email sends on signup
- [ ] Password reset email arrives
- [ ] Emails have correct content
- [ ] Links in email work
- [ ] Unsubscribe link works

Tell AI:

Add test mode checking:
Log when using test API keys.
Show banner: "TEST MODE - No real charges"
Use Stripe test cards only in development.

Finding Bugs Yourself

How to break your app:

1. Click fast and repeatedly

• Double-click submit button
• Click back button quickly
• Refresh during loading

2. Enter unexpected data

• Copy/paste 10,000 characters
• Enter emojis 🎉 in text fields
• Try SQL: '; DROP TABLE users;--
• Enter HTML: <script>alert('xss')</script>

3. Change state unexpectedly

• Log out in another tab
• Let session expire, then try action
• Open same page in multiple tabs

4. Use slow connection

• Chrome DevTools → Network → Slow 3G
• Try all main actions
• Do loading states show?

See BREAKING-THINGS.md for full list.

Documenting Bugs

Bug report template:

Bug: [Short description]

Steps to reproduce:
1. [First action]
2. [Second action]
3. [What triggers the bug]

Expected: [What should happen]
Actual: [What actually happened]

Screenshot: [Attach if visual]
Browser: [Chrome 120 on Mac]
URL: [Where it happened]
Frequency: [Always / Sometimes / Once]

Give to AI to fix:

Bug found: [paste bug report]

Please:
1. Reproduce the bug
2. Identify root cause
3. Fix the issue
4. Verify fix works
5. Test that fix didn't break anything else

Regression Testing

After fixing a bug, test:

Regression Checklist:
- [ ] Original bug is fixed
- [ ] Happy path still works
- [ ] Related features still work
- [ ] No new errors in console
- [ ] No new visual issues

Common regression issues:

• Fix breaks different browser
• Fix breaks mobile layout
• Fix breaks related feature
• Fix introduces new edge case bug

Pre-Deployment Checklist

Before pushing to production:

Production Readiness:
- [ ] All features tested (happy path + edge cases)
- [ ] Works on mobile (iPhone + Android)
- [ ] Works in Safari and Chrome
- [ ] No console errors
- [ ] Forms validate correctly
- [ ] Authentication works
- [ ] No test data visible
- [ ] Error messages are user-friendly
- [ ] Loading states show
- [ ] Can't break it with weird input

When to Get QA Help

Consider hiring QA when:

• 10 features to test before launch
• Complex user flows (multi-step processes)
• Multiple integrations to verify
• Preparing for big launch (> 1000 users)

For most MVPs: Following this checklist is sufficient.

Common Testing Mistakes

Quick Testing Shortcuts

5-minute quick test:

1. Load page in Chrome
2. Complete main user flow
3. Try one edge case (empty field)
4. Check mobile view (Chrome DevTools)
5. Look for console errors

15-minute thorough test:

1. Happy path in Chrome
2. 3-4 edge cases
3. Test on real phone
4. Check Safari
5. Try to break it (fast clicks, weird input)
6. Check all error messages

Testing Tools

Built into browser:

• Chrome DevTools (Inspect)
• Network tab (check API calls)
• Console (check for errors)
• Device mode (test mobile sizes)
• Lighthouse (performance + best practices)

Free external tools:

• BrowserStack (free trial) - Test on real devices
• Can I Use (caniuse.com) - Check browser support
• Validator.nu - Check HTML validity

Usually don't need:

• Automated testing frameworks (too complex for non-technical)
• Paid testing services (manual testing sufficient for MVP)

Success Looks Like

✅ Main user flow works perfectly ✅ Edge cases handled gracefully ✅ Works on mobile and desktop ✅ Works in Chrome and Safari ✅ Error messages clear and helpful ✅ Can't break it with weird input ✅ No bugs reported by users