Skip to content

v1.0.1 Refactoring Report

Scope

This audit reviewed every runtime package: application, domain, infrastructure, presentation, and utils, plus packaging, tests, examples, scripts, and documentation.

Applied non-breaking improvements

  • Removed unused runtime dependencies: numpy, pandas, plotly, and pydantic-settings.
  • Deduplicated requested report and chart export formats while preserving first-requested order.
  • Added validation for invalid SHA-256 read chunk sizes, preventing an infinite read loop for zero or negative values.
  • Corrected the test fixture's incorrect SHA-256 expectation and added regression coverage.
  • Escaped chart HTML and SVG text and safely serialized inline chart JSON.

Deferred to a future major release

StatisticsEngine and StatisticalAggregator duplicate percentile interpolation and basic-summary work, but expose different public keys and error wording. Consolidating them now would risk downstream compatibility. Extract a shared internal numeric-summary primitive in v2 with deprecation coverage.

The benchmark orchestrator remains intentionally broad because it owns lifecycle, retries, persistence, metrics, and report hooks. Split it only with an explicit public composition-root migration plan.