Skip to content

Triage the oracle's 10 failing and 13 unverified criteria without claiming a compliance tier #51

Open

Description

@hyperpolymath

The repaired oracle on PR #50, exact head b2edb23d84f9f61929e803e1b4a3e228967810b4, now executes successfully in Dogfood run 43832977797. Successful execution is not a compliance certificate.

The actual output is:

RSR v2.0 scorecard — subject (spec 2.1.0-draft)
tier=none score=72.11% coverage=71% provisional=true profile=true
verdicts: pass=25 partial=0 fail=10 unverified=13 na=30
failing (applicable): 1.2.4, 2.1.4, 2.1.5, 2.2.1, 2.3.1,
3.1.2, 3.1.3, 3.1.4, 3.1.8, 3.2.1

Evidence boundary:

  • Oracle Hypatia source: 51539abf3d2472b955da9b051a68c5971f6cfa22.
  • Runtime: Elixir 1.19.4 / OTP 27.3.4.19.
  • Retained standards source: e064539116c2dcfdf282a4749dbd58c69961c7e6; CLI reads 0-canon/rsr/rsr-criteria-v2.a2ml from that historical source.
  • The workflow preserves the existing score command and failure behavior. It returns successfully with this provisional result; the job's green status does not assert that every criterion passed.
  • The pin advances the scoring implementation as well as adding the compiler guard. Do not equate its score with a differently pinned scorer without a controlled comparison.

Follow-up:

  1. Obtain per-criterion evidence with these exact inputs, including all 13 unverified rows.
  2. Separate real repository gaps from stale source/layout assumptions or unsupported evidence checks. Fix a bad checker at its owning source; preserve genuine failed obligations.
  3. Reconcile duplicate work with the Pages/badge migration Migrate Pages and K9 badge publication together to the current template contract #49, historical replay Reconcile pre-mint answer handling with post-mint placeholder rejection #23 and the existing repository identity/map work repo: un-minted (template README) but referenced by epistemic-types#32 and occupancy-types — define, mint, and register in the type-family map #2.
  4. Repair bounded confirmed gaps, rerun positive and negative controls, and retain an exact-source receipt.
  5. Report the final census and provisional/tier state honestly. Do not relabel unverified rows as passing or weaken the applicability profile merely to gain a tier.

Stable canon publication identity and its separate normative-oracle evidence remain upstream work. This issue records the newly available score; it does not assert that all failures were introduced by PR #50.

Consistent-pin rerun — 2026-11-16

Final PR head 83363da45e4bfcad136dc73b56c7e8f3f1f1cbef aligns oracle, scanner and reports on Hypatia 6693f9aed8e4383f78a3d01b600b32663bf7a45c. Run 43838548141 completed the oracle successfully with the same census above: 25 pass, 10 fail, 13 unverified, 30 not applicable, provisional and no tier. Runtime and historical standards input remain as recorded. This establishes execution and pin consistency; the criterion follow-up remains open.

Metadata

Metadata

Assignees

No one assigned

    Labels

    bugSomething is broken or behaves incorrectlygovernancePolicy, rulesets, standards, compliance, and their enforcement

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions