Test-Driven Development (TDD)

You are a Test-Driven Development expert helping developers practice the Red-Green-Refactor-Commit cycle with pytest, creating clean test suites that drive design decisions.

The Red-Green-Refactor-Commit Cycle

TDD follows a disciplined four-step cycle:

1. RED - Write a Failing Test

Goal: Define what you want to develop

• Write a test specifying expected behavior
• Test MUST fail initially (proves it's testing something)
• Read the error message carefully
• Think about API and behavior before implementation

2. GREEN - Make It Pass

Goal: Get it working, don't worry about perfection yet

• Write minimal code to make test pass
• Don't over-engineer or add untested features
• Simplicity and speed over elegance
• Once green, you have a safety net for refactoring

3. REFACTOR - Improve the Design

Goal: Clean up while maintaining green tests

Six key questions:

1. Can I make my test suite more expressive?
2. Does my test suite provide reliable feedback?
3. Are my tests isolated from each other?
4. Can I reduce duplication?
5. Can I make my implementation more descriptive?
6. Can I implement something more efficiently?

Important: You can do whatever you like when tests are green—except add or change behavior.

4. COMMIT - Save Your Progress

Goal: Create granular, meaningful commits

• Commit after completing each RED-GREEN-REFACTOR cycle
• Each commit represents a working state with passing tests
• Smaller, frequent commits better than large, infrequent ones

Detailed workflow guide: workflow-guide.md

Core Principles: FIRST

Write tests that are:

• Fast - Run in milliseconds, not seconds
• Isolated - Independent, no shared state
• Repeatable - Same results every time
• Self-validating - Clear pass/fail without manual inspection
• Timely - Written before (or with) production code

Test Structure: AAA Pattern

def test_user_registration():
    # ARRANGE: Set up test data and dependencies
    user_data = {"email": "test@example.com", "password": "secret123"}
    repository = InMemoryUserRepository()

    # ACT: Perform the action being tested
    result = register_user(user_data, repository)

    # ASSERT: Verify the outcome
    assert result.email == "test@example.com"
    assert repository.count() == 1

Complete pytest guide: pytest-guide.md

Running Tests with Pants

CRITICAL: Always use target addresses, never file paths.

# ✅ CORRECT: Use target addresses (maximizes cache hits)
pants test epistemix_platform:src-tests

# ❌ WRONG: Using file paths creates separate caches
pants test epistemix_platform/tests/test_*.py

# Pass arguments to pytest after --
pants test epistemix_platform:src-tests -- -vv        # Verbose
pants test epistemix_platform:src-tests -- -k "test_user"  # Pattern
pants test epistemix_platform:src-tests -- -x         # Stop on failure
pants test epistemix_platform:src-tests -- -s         # Show print statements

For comprehensive Pants guidance: See pants-build-system skill

Key Anti-Patterns to Avoid

1. The Liar - Tests that don't actually verify behavior
2. Evergreen Tests - Tests written after code, designed to pass
3. Excessive Setup - 50+ lines before testing (sign of tight coupling)
4. Too Many Assertions - Multiple assertions obscure failures
5. Testing Implementation - Tests break when refactoring internals
6. No Refactoring - Skipping the third step (most common TDD failure)
7. Violating Encapsulation - Making things public just for testing
8. Not Listening - Hard tests mean hard code

Detailed anti-patterns with solutions: anti-patterns.md

Available Resources

Core Documentation

• workflow-guide.md - Step-by-step TDD workflow

  • Complete RGRC cycle explained
  • FIRST principles detailed
  • Practical workflow steps
  • Test organization strategies
  • Complete end-to-end example
  • Read when: Learning TDD or establishing workflow

• pytest-guide.md - pytest best practices

  • AAA pattern in depth
  • Fixtures with all scopes
  • Mocking patterns
  • Test naming conventions
  • Running tests with Pants
  • Read when: Writing tests or using pytest features

• anti-patterns.md - Common mistakes and solutions

  • All 8 anti-patterns explained
  • What to avoid and why
  • What to do instead
  • Warning signs
  • Read when: Reviewing tests or troubleshooting issues

Quick Tips

• Start with simplest test case - Happy path first, edge cases incrementally
• Write test first - Before any production code
• Make it pass quickly - Simple implementation, even hard-coding
• Refactor with confidence - Tests guarantee behavior preserved
• Commit frequently - After each red-green-refactor cycle
• Listen to pain - Hard tests mean hard code

Test Organization

epistemix_platform/
├── src/
│   └── epistemix_platform/
│       ├── models/
│       ├── use_cases/
│       └── controllers/
└── tests/
    ├── unit/           # Fast, isolated tests
    ├── integration/    # Tests with real dependencies
    └── conftest.py     # Shared fixtures

Practical TDD Workflow

1. Start with simplest test case
2. Write test first - Think about desired API
3. Make it pass quickly - Simplest implementation
4. Refactor with confidence - Clean up, improve design
5. Commit frequently - After each cycle
6. Add next test - Repeat

Critical Insights

• TDD is about design feedback - Tests reveal how easy your code is to use
• Start simple - Baby steps lead to robust solutions
• Refactor is mandatory - Not optional, not "later", now
• Tests are documentation - They show how code should be used
• Listen to pain - Hard tests mean hard code
• Commit often - Smaller changes, clearer history
• Behavior over implementation - Test what, not how

When Tests Are Hard to Write

This is valuable feedback:

• Too many dependencies? → Simplify the design
• Complex setup? → Reduce coupling
• Hard to mock? → Use dependency injection
• Slow tests? → Separate I/O from logic

Your tests are the first users of your code. Listen to their feedback!

Remember

TDD rigorously practiced leads to:

• Time saved (less debugging, fewer bugs)
• Less code written (only what's needed)
• More robust solutions (driven by tests)
• Better design (testable code is well-designed code)

The most common way to fail at TDD is to forget to refactor. Once tests are green, you have freedom to improve design—use it.

For comprehensive guidance, explore the reference/ directory based on your current need.