testing

Testing Standards

Core Philosophy: Test Reality, Not Fiction

Why NO MOCKS? Mocks test your assumptions, not your code. Real bugs hide in integration points, not unit logic. Better approach: No test is better than a false-confidence mock test.

Core Principle: NO MOCKS, NO FAKE DATA

STRICT RULE: Never use mocks, stubs, or fake datasets. If real testing isn't possible, don't write tests.

• No mock objects - Use real implementations
• No mock datasets - Use actual sample data
• No stub services - Connect to real test instances
• Alternative: Ask user for sample data or test environment setup

When to Write Tests

• DO: Test with real data and actual dependencies
• DO: Use test databases with real schemas
• DO: Test against actual file systems
• DON'T: Write tests if only mocks would work
• DON'T: Create artificial test scenarios

Test Structure

tests/
  conftest.py          # Real test fixtures
  sample_data/         # Actual data samples (user-provided)
    valid/
    invalid/
  integration/         # Tests with real dependencies
    test_database.py   # Real DB connection
    test_api.py        # Real API calls

Frameworks (Language-Specific)

• Python: pytest with real fixtures
• JavaScript: vitest or jest (no mocking libs)
• Database: Use test DB with real migrations
• APIs: Test against staging/local instances

Writing Real Tests (Think About Value)

# GOOD: Tests actual behavior
def test_user_creation(real_db):
    """Tests that users are actually persisted."""
    user = User.create(email="[email protected]")
    # This catches: ORM issues, DB constraints, connection problems
    assert real_db.query(User).filter_by(email="[email protected]").first()

# BAD: Tests nothing meaningful
# def test_user_creation(mock_db):  # NO!
#     mock_db.return_value = User()  # Tests that Python works?

Ask: What am I actually testing? Would this catch real bugs?