benchmark_harness_plus 3.0.1

A statistically rigorous benchmarking harness with median-based comparisons, CV% reliability metrics, and outlier-resistant measurements.

3.0.1

• Fix typo in README

3.0.0

• Breaking: Added Percentage type for type-safe percentage values
• Breaking: cv() now returns Percentage instead of double
• Breaking: BenchmarkResult.cv now returns Percentage
• Breaking: BenchmarkResult.improvementVs() now returns Percentage
• Breaking: BenchmarkComparison.improvementPercent renamed to improvement and returns Percentage
• Breaking: reliabilityFromCV() now takes Percentage instead of double

2.0.0

• Bump SDK constraint to ^3.5.0
• Update extra_pedantic to ^5.0.0
• Add GitHub Actions CI

1.2.0

• Removed GC triggering between variants (can cause more noise than it reduces)

1.1.0

• Added "fastest" column to table output showing the minimum (best) sample time
• Added "Fastest" line to detailed result output
• Added new example demonstrating when minimum vs median should be used
• Restructured examples into separate files for easier reference

1.0.0

• Initial release
• Statistical functions: mean, median, stdDev, cv, min, max
• BenchmarkConfig with quick, standard, and thorough presets
• BenchmarkResult with computed statistics and reliability assessment
• BenchmarkComparison for comparing variants
• Benchmark harness with warmup and randomized ordering
• Result formatting: table, detailed, and CSV export
• Comprehensive documentation and examples