Optimize AssemblyNameViolationDetector to O(n) time complexity

[mfogliatto]

Description

This PR improves the time complexity of AssemblyNameViolationDetector from O(m*n) to O(n), where m is the number of rules and n is the number of references. The optimization is achieved by reversing the lookup strategy: instead of iterating over rules for each reference, we now build a dictionary of references and check each rule against it in O(1) time.

Type of change

• Enhancement (e.g. new feature, functional improvement, etc.)
• Bug fix
• Chore (e.g. refactoring, documentation, etc.)

Related Issue

Closes #28

Implementation Details

Core Changes

• Optimized AssemblyNameViolationDetector: Added a new experimental implementation that builds a dictionary from references for O(1) lookups
• Configuration Support: Added UseExperimentalDetectors and EnableDebugMessages flags to ReferenceCopConfig to enable the new implementation
• Backward Compatibility: The original implementation remains available; users can opt-in to the experimental version via config

Performance Improvements

Benchmark Results

Performance testing was conducted using BenchmarkDotNet across various scenarios with different numbers of rules and references:

Scenario	Rules	References	Original (μs)	Experimental (μs)	Speedup	Memory Reduction
Small	10	10	6.513	5.262	1.24x	15% less
Medium	50	100	301.405	134.393	2.24x	2% less
Large	100	500	3,141.434	1,153.985	2.72x	~1% less

Key Findings:

• Up to 2.72x faster for large codebases (100 rules, 500 references)
• Consistent speedup ranging from 1.24x to 2.72x across all test scenarios
• Reduced memory allocations by up to 15% in small scenarios
• Better scaling characteristics as the number of rules increases

The experimental implementation shows the most significant improvements when dealing with:

• Large number of rules (50-100+)
• Moderate to large number of references (100-500+)
• Scenarios typical of enterprise codebases

How Has This Been Tested?

• Validated in MSBuild Task integration
• Validated in Roslyn Analyzer integration
• Tested using the playground project with real-world scenarios

Screenshots (if applicable)

N/A - Performance improvement, no UI changes

Additional context

• The implementation maintains full backward compatibility
• The original algorithm remains the default to ensure stability
• Users can opt-in to the optimized version
• Future work may include making the experimental version the default after additional validation
• Changes include playground project improvements to facilitate e2e testing and validation