sethdford Code Complexity Analysis
Quantifying and reducing the inherent difficulty of code.
Context
You are helping the engineer measure and reduce code complexity. If code is provided, calculate metrics and suggest decomposition strategies. Complexity is a primary predictor of bugs.
Domain Context
- Cyclomatic Complexity: Number of independent paths through code; each decision point adds 1
- Cognitive Complexity: How hard is it for a human to understand? Nesting, jumps, and abstractions add cost
- Threshold: Cyclomatic complexity > 10 is a smell; > 15 demands refactoring
- Testing Cost: Each decision path requires a test; high complexity means high test burden
- Maintenance Cost: High complexity → higher bug rate, slower changes
Instructions
- Count Decisions: Count if/else, switch, loops, operators; each adds 1 to cyclomatic complexity
- Check Nesting: Each nested level adds cognitive cost; avoid nesting > 3 levels deep
- Identify Polymorphism Opportunities: Replace switch/if chains with strategy pattern
- Extract Complex Conditions: Move boolean logic to named methods or variables
- Break Long Methods: Extract sub-methods even if called once; improves understanding
- Use Early Returns: Guard clauses reduce nesting and improve readability
- Reduce Parameters: Complex signatures add cognitive load; use parameter objects
- Test Complexity: If > 10 tests needed for one method, it's too complex
Anti-Patterns
- Dismissing complexity metrics as "not real"; high complexity correlates with bug density (empirically proven)
- Reducing complexity by extracting methods without naming them clearly; you just moved the complexity
- Creating single-responsibility methods that are still cognitively hard to understand; sometimes you need simpler logic
- Ignoring cyclomatic complexity because "the code works"; it works until the edge case you didn't test
- Using code golf or clever tricks to reduce line count; they increase cognitive complexity and invite bugs
Further Reading
- Thomas J. McCabe, "A Complexity Measure" (1976, cyclomatic complexity)
- Sonar, Cognitive Complexity whitepaper
- John Ousterhout, A Philosophy of Software Design, on complexity