Skip to content

fix(lineage): bind dynamic evaluations to substantive criteria - #915

Draft
seonghobae wants to merge 50 commits into
feat/dichotomous-measurement-policyfrom
feat/dynamic-evaluation-lineage-20260902
Draft

fix(lineage): bind dynamic evaluations to substantive criteria#915
seonghobae wants to merge 50 commits into
feat/dichotomous-measurement-policyfrom
feat/dynamic-evaluation-lineage-20260902

Conversation

@seonghobae

@seonghobae seonghobae commented Sep 2, 2026

Copy link
Copy Markdown
Contributor

Root-cause repair

Dynamic-evaluation lineage preserves what was evaluated and keeps criterion, observation, adjudication, calibration, promotion, linking, and supersession provenance auditable without absorbing foreign scoring or provider authority.

Current behavior

  • criterion-bound runs retain one complete immutable criterion-set snapshot and reject mixed set/digest/rubric/blueprint semantics;
  • direct item construction type-checks criterion_refs before truthiness-based binding inference;
  • zero-anchor unavailable and within_run_only runs remain valid during cold start;
  • calibration, promotion and linking evidence identities remain distinct;
  • adjudication case/resolution identities remain distinct and source observations are never overwritten;
  • in-run supersession is acyclic and bounded;
  • opaque provenance references reject padding, controls/surrogates, Unicode Cf, Zl, Zp, and the 66 Unicode noncharacters without normalizing identity.

Owner boundary

LineageWeave owns the provenance/read-model projection only. contextual-orchestrator owns provider/model execution and invocation evidence; fast-mlsirm owns reusable psychometric calibration/linking arithmetic; TEPP owns temporal/drift semantics. Production integration must consume released/versioned owner artifacts and exact digests. No published fast_mlsirm_dynamic_evaluation_item/v1 contract is invented; tests use synthetic_source_contract/v1. ADR 0355 remains Proposed.

TDD repair lineage

  • RED 6e83d1f8e151f91c442e20e42d28a5fef8118b46 → GREEN 1061fb316228968ece433d7f8c8848b200af7e17: calibration/promotion identity separation;
  • RED 8c73220d49f429f2ce2930438fb63a5b672b246e → GREEN 9683ca64e87940d1fc7b878e4b89dd0511c2ab5d: linking-evidence identity separation;
  • RED 9a78c0a2e5e7546d8211d008b699ce1e4f9412be → GREEN 3066c12eacbe86977f55b1abf3ebcf9f05e39b7a: malformed falsy criterion_refs fail the collection boundary;
  • 05ce15a3f57254e73db2fd00f2e574abdf8d9f97: synthetic fixture no longer names a nonexistent fast-mlsirm contract;
  • RED 79c1c32a0af9d45940b8dca12a13b7af03b89271 → GREEN 290da3a20ff76ae1a32b50166ed34bfd1679a5c7: reject line/paragraph separators;
  • RED 623583464925d404f05750c841a4ca2f3b9cdeb8 → GREEN 0e1a413fd92e14f289ac648a75d621d26f0d3cb4: reject Unicode noncharacters while preserving ordinary unassigned code points; e977ba3f14b799b75b88467f2c25c04424e7299d and 958e68de450bfa2ff0a3aef54716a10ea4ce7976 keep changelog/ADR evidence current.

Current stack / merge boundary

Parent #902 moved non-force to c316bfbde8644feda7e49579d43cc162e99b0bad after #899/current-main convergence. Review 5118128121 verified this PR's 12 dynamic-evaluation lineage/evidence files are disjoint from the inherited workflow/contract repair. Two-parent convergence 8c74712f2af4acf76871c71a9fd72c6aa2912f4a preserves the lineage tree, adopts the moved parent's protected workflow contract, and joins #902 without force rewrite.

  • exact parent/base: #902@c316bfbde8644feda7e49579d43cc162e99b0bad;
  • exact current head: 8c74712f2af4acf76871c71a9fd72c6aa2912f4a;
  • Draft/open/mechanically mergeable;
  • exact-head Tests 33923173150 completed skipped under Draft admission and is not product GREEN;
  • predecessor checks/reviews do not transfer and no qualifying independent approval is claimed.

Keep Draft until current-head repository/security/governance evidence is available and terminal-clean and a qualifying independent approval applies. No self-approval, force-push, bypass, fabricated owner contract, or predecessor evidence transfer.

@coderabbitai

coderabbitai Bot commented Sep 2, 2026

Copy link
Copy Markdown

Important

Draft PR not reviewed

Draft PRs are not automatically reviewed by default.

  • Trigger a manual review

To automatically review draft PRs, update your CodeRabbit configuration:

reviews:
  auto_review:
    drafts: true

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

Comment @coderabbitai help to get the list of available commands.

@seonghobae seonghobae changed the title feat(lineage): project dynamic evaluation snapshots fix(lineage): bind dynamic evaluations to substantive criteria Sep 2, 2026
Adopt #9028a8b269 after its protected-main convergence. Preserve the dynamic-evaluation lineage/evidence delta and the current Tests workflow without rewriting branch history.
Adopt #9022dd6258 after protected main #924 while preserving the dynamic-evaluation lineage/evidence delta without force rewrite.

@seonghobae seonghobae left a comment

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Parent #902 has now moved non-force to c316bfbd… after #899/current-main convergence. This branch’s 12 dynamic-evaluation lineage/evidence files are disjoint from the inherited workflow/contract repair. Keeping the old #902 ancestry is therefore a wrong-base finding, not a reason to close or rewrite history. Preserve the lineage delta exactly, adopt the moved parent’s protected workflow contract, and advance with a two-parent non-force convergence commit; predecessor checks/reviews remain non-transferable.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant