Skip to content

Reject ambiguous bootstrap count rewrites #461

Closed

Description

@afflom

Follow-up to #442 and its source review: #442 (review). Preserve #442's ownership and fast/report design; this is a narrow repair in its existing planner.

Two source-traced counterexamples at 0eff75c:

  1. number_map ignores unchanged pair explanations. A 10→11, B 20→20, C 12→12 and D 18→18 explain old 30 as both A+B→31 and C+D→30. The writer must flag the conflict rather than rewrite to 31.
  2. Prose matching selects the first old-number/context anchor. Two matching anchors for 30 with destinations 31 and 32 must be flagged regardless of case order; the conflict must not fall through to shared-number fallback.

Acceptance: focused tests calling number_map and prose cover unchanged and moved pair collisions, agreeing destinations, unchanged sums, overlapping context matches, order independence and fallback suppression. Keep default conformance assertions, seeds/bounds, model claims and generated products unchanged. Native tests, formatting and Clippy must qualify the repair; source/reference-model checks alone are not Rust execution. Integrate through a draft stacked PR based on #442, without rewriting its contributor branch.

Activity

  1. afflom commented on Nov 13, 2026

    @afflom
    MemberAuthor

    Delivery audit: #462 was externally merged as6a02d2c7d519f1a4d169ebb3496fbad51bd259cb, ordered parents acceptedbe7d2f8a + reviewed810cf9df. The actual net change is only xtask/src/counts.rs, byte-identical to the reviewed/native-tested repair. The full issue's focused counterexamples, unchanged products/seeds/model, native tests, format and Clippy retain their existing evidence; no checks were repeated.

    Keep this issue open while the actual merged tree finishes its required qualification. Push43494150955 currently has4 successful,12 active and39 queued jobs; exact repro43494150940 and same-run honesty passed. The old3-success/7-skip development run does not qualify this composed tree, and no final premerge acceptance review was found. Once this existing actual-tree run completes and its acceptance is reviewed, explicitly close with #462's merged link. This is an acceptance gap after an external landing, not permission for another producer or automatic rerun.

  2. afflom commented on Nov 14, 2026

    @afflom
    MemberAuthor

    Delivered through merged #462. Actual6a02d2c7d519f1a4d169ebb3496fbad51bd259cb/tree d3dd48a52f369afd7627df01e51ebeaef10eeaeb passed its full70-job/772-step CI43494150955, same-run honesty and exact repro43494150940. Independent final acceptance confirms both planner counterexamples, all requested focused coverage, native tests/formatting/Clippy and unchanged products/seeds/model scope. The actual counts.rs equals the reviewed repair; the qualified tree is the actual merged tree.

    Closing this narrow issue as completed after postlanding review. This does not retroactively claim premerge acceptance or complete broad roadmap parents. Existing evidence was reused without repeating checks or producers.

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

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions