Skip to content

Add performance budget regression guardrails - #60

Merged
josiahking merged 1 commit into
2.xfrom
feature/phase-6-5d-performance-budgets
Sep 2, 2026
Merged

Add performance budget regression guardrails#60
josiahking merged 1 commit into
2.xfrom
feature/phase-6-5d-performance-budgets

Conversation

@josiahking

Copy link
Copy Markdown
Owner

Summary

  • add benchmark-only performance budget evaluation with pass/warn/fail/incomparable states
  • add tracked regression calibration and reference artifacts for the accepted PHP 8.4.25 benchmark lane
  • enforce evidence identity, cleanliness, source SHA, sample protocol, timing units, and normalized artifact integrity
  • add a CI-safe benchmark policy job without running authoritative wall-clock timing on shared runners
  • strengthen architecture/documentation guardrails for benchmark policy and CI workflow structure

Validation

  • deterministic package split validation passed for all 7 packages
  • release package validation passed
  • skeleton create-project validation passed
  • prerelease consumer validation passed
  • benchmark policy passed: 22 tests, 139 assertions
  • root quality passed: Deptrac, PHPStan, PHP-CS-Fixer, PHPUnit
  • root PHPUnit: 679 tests, 2456 assertions, 1 skipped
  • Composer audit: no security advisories
  • license policy passed

Notes

  • warm HTTP p50 budgets are EvolvePHP self-regression budgets, not competitor-relative rankings
  • application_boot remains monitor-only
  • shared GitHub-hosted runners validate policy/schema/smoke behavior only; authoritative absolute timing remains controlled evidence
  • this change does not modify production framework runtime behavior

@josiahking josiahking self-assigned this Sep 2, 2026
@josiahking
josiahking merged commit 7077c9b into 2.x Sep 2, 2026
4 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant