Skip to content

test(testing): pin a negative remedy assertion for rule-throw-only-oracle #6386

Closed

Description

@kyle-sexton

Summary

Pin a negative assertion on the remedy text of rule-throw-only-oracle in the cannot-fail scan tests.

Context

In cant-fail-scan.test.sh the rule's remedy text has two positive pins but no negative one, unlike its peer rules. The detector-findings README adopter tally therefore reads 7 of 13.

Done when

  • cant-fail-scan.test.sh has an assert_not_contains that the throw-only remedy is not the weak-oracle remedy ("Assert the value the code should produce").
  • The detector-findings README adopter tally reads 8 of 13.

Activity

  1. added
    needs-triageNot yet classified. Floor until a type and one priority tier are set.
    on Oct 4, 2026
  2. github-actions commented on Oct 4, 2026

    @github-actions
    Contributor

    Thanks for the report. I verified this against the repo: cant-fail-scan.test.sh does pin two positive assertions for rule-throw-only-oracle's remedy but no negative one, unlike its peer report-only rules, and the detector-findings README's remedy-assertion tally for testing:audit indeed reads 7 of 13 for this reason. The requested fix — adding an assert_not_contains confirming the throw-only remedy text isn't the weak-oracle remedy text, then updating the tally to 8 of 13 — is clear, small, and self-contained with no open questions. I didn't find any existing open issue covering this specific gap (related issues #6040, #6038, #6042 are broader audit feature work, not this assertion).

    Automated triage. A maintainer reviews it.

  3. added
    good first issueGood entry point for new contributors.
    priority: lowNice-to-have, cosmetic, or speculative; opportunistic.
    on Oct 4, 2026
  4. kyle-sexton commented on Oct 11, 2026

    @kyle-sexton
    MemberAuthor

    Draft PR: #7181 (branch test/6386-throw-only-negative-remedy).

    • Code state: complete. cant-fail-scan.test.sh now asserts that the throw-only remedy is not the weak-oracle remedy. The detector-findings testing:audit row reads 8 of 13, and convention changelog entry 3.7.1 records the change.
    • Verification is partial. The new assertion passes, but the full cant-fail-scan.test.sh run was stopped when the overnight run wound down (1,283 PASS, 0 FAIL at that point). The fragment check and markdownlint pass. affected-tests.sh --run was not run.
    • Next step: run scripts/affected-tests.sh --base origin/main --run on the branch, or mark the draft ready so CI's test-bash lane runs the suite. Marking it ready is Kyle's call.

    Generated by Claude Code

  5. removed
    needs-triageNot yet classified. Floor until a type and one priority tier are set.
    on Oct 12, 2026
  6. added theissue type on Oct 12, 2026
  7. kyle-sexton commented on Oct 13, 2026

    @kyle-sexton
    MemberAuthor

    Decision recorded. The owner approved going with independently verified recommendations on 2026-10-13. Two researchers worked this issue separately (code and history; upstream docs and owner intent), and a judge reconciled them.

    • Is 'only a full suite run remains' on draft PR test(testing): pin a negative remedy assertion for rule-throw-only-oracle #7181 true? → No. test(testing): pin a negative remedy assertion for rule-throw-only-oracle #7181 is CONFLICTING with main in docs/conventions/detector-findings/CHANGELOG.md, because main's top entry is now [3.8.0] - 2026-10-10 and the PR adds 3.7.1. cant-fail-scan.test.sh and README.md merge cleanly. Remaining work: merge main, renumber the entry 3.8.1 and place it above 3.8.0, run the full cant-fail-scan.test.sh and affected-tests, then flip to ready.
      Basis: gh pr view 7181 (isDraft true, mergeable CONFLICTING); git merge-tree --write-tree origin/main origin/test/6386-throw-only-negative-remedy: CONFLICT only in docs/conventions/detector-findings/CHANGELOG.md; origin/main CHANGELOG.md first heading '## [3.8.0] - 2026-10-10'
    • Is the README tally change still needed? → Yes. Main's README still reads 'Remedies PARTLY pinned, 7 of 13', with rule-throw-only-oracle 'owed its negative'.
      Basis: origin/main:docs/conventions/detector-findings/README.md:737
    • Does feat(testing): judge Reqnroll [Then] steps as tests in the cannot-fail scan #7049 still collide with the test file? → No. The test file auto-merges, and the convention changelog is the only conflict.
      Basis: git merge-tree output this run ('Auto-merging plugins/testing/skills/audit/scripts/cant-fail-scan.test.sh')

    Recommendation: Have an agent finish #7181: merge main, resolve the changelog conflict by renumbering to 3.8.1 above 3.8.0, run the full cant-fail-scan.test.sh and affected-tests, and leave the PR as a draft. The ready flip stays a stop per AGENTS.md unless the task authorizes it.

    Where sources disagreed, decided by: No conflict. Both researchers agree and this run confirmed every fact on origin/main. Rule 1 (AGENTS.md) only keeps the ready flip as a stop.

    Admission to an agent lane stays with the work-class rules and owner ratification.

    This was generated by AI.

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

Metadata

Metadata

Assignees

No one assigned

    Labels

    good first issueGood entry point for new contributors.priority: lowNice-to-have, cosmetic, or speculative; opportunistic.

    Type

    Fields

    No fields configured for Task.

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions