test-data-management

Test Data Management

Strategies for creating and maintaining test data.

Context

You are designing test data infrastructure. Good test data is realistic but lightweight.

Domain Context

• Fixtures: Pre-built test data sets; inflexible but fast
• Builders: Fluent API to construct objects; flexible, readable
• Factories: Methods that generate test data; reusable, minimal
• Realistic: Use production-like data; but omit sensitive info
• Minimal: Only data relevant to the test; avoid clutter

Instructions

1. Choose Strategy: Fixtures (pre-built), Builders (flexible), Factories (reusable)
2. Use Builders for Complex Objects: Clear intent; easy to vary specific fields
3. Create Factories for Common Patterns: User factory, Order factory
4. Avoid Shared State: Each test gets fresh data; no pollution
5. Seed Data Carefully: Use realistic but non-sensitive data
6. Document Defaults: Default builder state should be valid for most tests
7. Performance: Consider test database resets vs. transactions for cleanup

Anti-Patterns

• Massive fixture files with hundreds of records; hard to understand which data matters
• Global shared test data; tests interfere with each other
• Over-realistic test data (full names, addresses); wastes resources
• Complex object construction in tests; use builders instead
• Test data stored in files separate from tests; couples tests to external state

Further Reading

• Gerard Meszaros, xUnit Test Patterns, on Test Fixture Strategies
• Roy Osherove, The Art of Unit Testing, on Test Data