Repository navigation
Benchmark Bifrost v0.13.0 for DataFlowBench v0.9.1 - #293
Merged
Merged
Conversation
DavidBakerEffendi
changed the title
7 tasks
DavidBakerEffendi
added a commit
that referenced
this pull request
Oct 26, 2026
Refresh DataFlowBench's Bifrost evidence from v0.12.0 to the checksum-verified public v0.13.0 release against the unchanged 1,108-case v0.9.0 population. All 20 Bifrost partitions are freshly measured; the other 72 partitions retain their original v0.9.0 report and raw-evidence bytes. This PR is the evidence phase of v0.9.1; the release freeze and publication must follow its squash merge in a separate PR. **Key Changes**: - Bind the published crate, official macOS archive, executable digest/build identity, exact runner source and population; retain all first-attempt command streams, report bytes, mappings and checksummed receipts. The serial matrix records 1,116 fresh rows and 3,136 carried rows without changing expected behavior or denominators. - Update the exact v0.13.0 native capability rationale for the standalone engine's empty product-rule catalog. Preserve the failed strict qualification and failed pre-dispatch setup attempt. The independently recomputed control review permits measurement with three unchanged JavaScript inconclusive states and 15 unbound-endpoint controls; it explicitly does not claim full policy qualification or clean negatives. - Add an evidence verifier, integrity/identity regression tests, and a v0.9.1 publication checker that reuses the established clean-clone, official-main, tag and generated-results gates. Keep cold timing diagnostic; older warm and overhead evidence retains its original version provenance. **Touch Points**: - src/native.rs and src/tests/native.rs - scripts/execute-release-v091.py, scripts/qualify-bifrost-v013.py, scripts/review-bifrost-v013-controls.py and focused tests - scripts/check-v091-evidence.py, scripts/check-release-v091-results.py and release-gate tests - docs/v0.9.1-release-contract.md, docs/v0.9.1-publication.md and docs/releases/v0.9.1-*.md - reports/releases/v0.9.1, .gitattributes and .github/workflows/ci.yml Validation: 27 native Rust tests, 10 executor tests, six native/strict qualification harness tests, six retained-evidence integrity tests and four publication-gate tests; cargo fmt --check, git diff --check, locked offline single-job runner build, 1,108-case validation, and existing 92-report validation passed. The full new evidence audit verifies all staged/raw mappings, 1,108 frozen case/fixture identities and the 72 carried reports/raw references. All 20 groups completed; the 1,116 fresh outcomes are 332 reached, 328 not-reached, 371 inconclusive and 85 unsupported, with zero per-case transitions from v0.12.0. Exact-head CI state is reported separately by GitHub checks. No v0.9.1 release freeze, tag, release or Pages deployment is claimed by this evidence PR. The publication runbook explains the remaining merged-main freeze and generated-artifact phase.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Refresh DataFlowBench's Bifrost evidence from v0.12.0 to the checksum-verified public v0.13.0 release against the unchanged 1,108-case v0.9.0 population. All 20 Bifrost partitions are freshly measured; the other 72 partitions retain their original v0.9.0 report and raw-evidence bytes. This PR is the evidence phase of v0.9.1; the release freeze and publication must follow its squash merge in a separate PR.
Key Changes:
Touch Points:
Validation: 27 native Rust tests, 10 executor tests, six native/strict qualification harness tests, six retained-evidence integrity tests and four publication-gate tests; cargo fmt --check, git diff --check, locked offline single-job runner build, 1,108-case validation, and existing 92-report validation passed. The full new evidence audit verifies all staged/raw mappings, 1,108 frozen case/fixture identities and the 72 carried reports/raw references. All 20 groups completed; the 1,116 fresh outcomes are 332 reached, 328 not-reached, 371 inconclusive and 85 unsupported, with zero per-case transitions from v0.12.0. Exact-head CI state is reported separately by GitHub checks.
No v0.9.1 release freeze, tag, release or Pages deployment is claimed by this evidence PR. The publication runbook explains the remaining merged-main freeze and generated-artifact phase.