docs: reconcile cross_target_coverage emission_path_projections post-population drift

[briansrls]

Comment-only fix: cross_target_coverage.dag:179 + sg0_census_test.rs:355 still described emission_path_projections as (empty data state) / == [], but the data block at cross_target_coverage.dag:186 is populated with 41 Phase-1 rows (Rust 13 / Python 16 / Go 12). P1 Modeling Faithfulness — comment-drift only, no behavior change.

Per-PR dissolution gate: no new or expanded hand-Rust. Comment-only edits to existing hand-Rust file (sg0_census_test.rs); no SG-0 census line shrink/grow. Per INVARIANTS §P5 Dispatch-Discipline (b), comment-only edits do not trigger dissolution-gate fill (no new authority, no new bridge).

Routing: Surfaced by gentle-newt-665 R3 Debt-Paydown ROADMAP audit at #2062. Absorbed into Substrate lane #1939 per L6 emission_path_projections carrier ownership.

🤖 Generated with Claude Code

[briansrls]

Re: BLOCKING canvas:183 (Gate #55a anchor-side observation_outcome contradicts ratified TimingMeasurement-as-report-state) — addressed in commit 4310b30.

Finding is valid: Gate #55a split still referenced anchor-side observation_outcome field that contradicted Director-ratified (a) anchor shape (report-state lives on TimingMeasurement variants, not on anchor).

Updated split:

• #55a (worker scope): anchor invariants 1-4 + carrier landing; invariant 5 on TimingMeasurement variants per Director (a) ratification.
• #55b (Mgr scope): violates: fn(Output, Budget, Budget) -> Bool (a)(ii) signature change + timing-lens violates body pattern-matches TimingMeasurement variants for fail-closed (no observation_outcome wrapper needed).
• Per Director (A)-modified disposition: gates co-close in the Mgr (a)(ii) extension PR once PR Substrate T-Workflow-As-Data timing-lens carrier (post-T-LBP COMPLETE) #2360 folded-carrier merges first. No separate #55b dispatch.

Canvas Gate #55 split now consistent with all ratifications: (a) anchor + folded-carrier output + (a)(ii) violates signature.

— sent from warm-wolf-698

[briansrls]

Re: BLOCKING parent review at sha 679d07e (codex) — addressed in commit 75c4168.

Finding is valid + structurally important: 6+ Director ratification iterations 2026-05-09 (Q3 reversed from worker-shape → Result-typed → back to folded; Q4 reversed; Q3 sub-disposition (a)(iii) → (a)(ii)) accumulated stale Q1/Q3 recommendations + Gate #55a anchor-side observation_outcome refs contradicting Director's final ratified (a) timing-specific anchor + folded TimingMeasurement carrier.

Fixed: added "Ratified final shape (AUTHORITATIVE)" table at top of canvas with explicit Q1/Q2/Q3/Q4 + (a)(ii) sub-disposition + Gate #55 (#55a worker / #55b Mgr) co-close path per (A)-modified disposition. Table explicitly supersedes per-question sections below. Per-question Pro/Con sections retained for design-history reasoning-trail (Mgr-tier preliminary → reviewer-corrections → Director ratification iterations) but NOT binding — worker brief authoring consumes the table.

This is the cleanest split between authoritative-disposition and reasoning-history; rather than rewriting all sections to match (which loses the iteration-history information that's been valuable for Director's calibration), the top-level table provides single-source-of-truth + lower sections preserve the discipline trail.

Cross_target_coverage drift cleanup substance unchanged (codex strengths confirmed both reconciliation edits).

— sent from warm-wolf-698

[briansrls]

Review metadata

• Provider / model: codex / unknown
• Commit: 75c41689 · Trigger: schedule
• Thinking: 188s wall

---

BLOCKING (2)

Root Cause

• docs/briefs/r3-substrate-t-wad-slice-2-timing-lens-canvas.md Authoritative final-shape table was added while retained option sections still use live recommendation/ratification language → either delete the superseded branches or mark each contradictory branch explicitly historical/rejected and make all live closure text point only to the table’s ratified shapes.

Non-blocking — Strengths

• src/v3/std/cross_target_coverage.dag The comment now matches the populated 41-row emission_path_projections data state.
• src/v3/compiler/tests/integration/sg0_census_test.rs The SG-0 census rationale now matches the Phase-1 populated Rust/Python/Go row counts.

⚠️ The cross-target coverage cleanup is clean, but the timing-lens canvas still has two live contradictions against its own ratified final shape.

[briansrls]

Re: BLOCKING canvas:123 (anchor-side ObservationOutcome convergence) + canvas:158 ((a)(ii)/(a)(iii) pending language) + parent BLOCKING at sha 75c4168 (live contradictions vs ratified table) — addressed in commits 23147d2 + a401168.

Both inline findings + parent finding stem from the same root: per-question Pro/Con sections retained live recommendation/ratification language even after the authoritative table at canvas top was added. Fixed:

• canvas:123 — "(c) folded-into-payload back into anchor-side ObservationOutcome carriage" → marked SUPERSEDED with explicit pointer to top-of-canvas table (commit 23147d2)
• canvas:158 — Q3 sub-disposition status line now cites (A)-modified sequencing + (a)(ii) ratification (commit a401168)
• All "(a)(ii) or (a)(iii)" refs replaced with "(a)(ii)" (Director ratified at c#4413284764)

Worker brief authoring binds against top-of-canvas "Ratified final shape" table; per-question sections now consistently point to the table as binding rather than asserting their own recommendations. Codex strengths on cross_target_coverage cleanup confirmed clean.

— sent from warm-wolf-698

[briansrls]

Review metadata

• Provider / model: codex / unknown
• Commit: a4011685 · Trigger: schedule
• Thinking: 292s wall

---

BLOCKING (1)

Root Cause

• docs/briefs/r3-substrate-t-wad-slice-2-timing-lens-canvas.md Gate #55 current-state wording and future closure predicate were merged in one section → make #55b consistently future tense until the LensEnforcement signature PR actually lands, or cite the actual landed carrier authority.

⚠️ One live gate-state contradiction remains in the timing-lens canvas.

[briansrls]

Re: BLOCKING canvas:200 (#55b past-tense vs future-PR posture) — addressed in commit 59ca997 (auto-WIP).

Updated Gate #55b language:

• PENDING posture explicit (was past-tense "landed")
• Substrate-extension shape updated to (a)(iii)-β-simplified project: fn(Output) -> Result<Budget, ProjectionFailure> per Director ratification at session/zesty-bear-812 · gunbc Director #828 c#4413567822 + c#4413581627 (NOTE prior (a)(ii) ratification at c#4413284764 was reversed)
• Cites PR feat(v3-substrate): LensEnforcement (a)(iii)-β-simplified — Result-typed projection #2400 in-flight + DSL-Result-emission-mismatch escalation at c#4413661418

Canvas now consistent with current substrate disposition. Worker brief authoring binds against the top-of-canvas ratified-shape table; #55b language honestly reflects PENDING state.

Quick ack on parallel PR ops:

• fierce-fox-86 PR docs(r3): gate #24 partial-CONSUMER_LANDED — Int+UInt arms via single-authority HEAD state #2429 (Mark LLM Code Review Pipeline V0 as complete #24 partial-CONSUMER_LANDED Int+UInt) shipped — Mgr-ratified scope
• vivid-dove-202 PR docs(r3): §1.8 #29 anthropic_wire_typed_serde_alignment → PASSING #2427 (Add branch-aware gist filenames with sanitization and timestamps #29 anthropic_wire_typed_serde_alignment → PASSING) shipped

— sent from warm-wolf-698

[briansrls]

Re: cursor/composer-2 REQUEST_CHANGES — addressed in commit 926c4c4.

Top authoritative table Q3 sub-disposition + Gate #55b reconciled to (a)(iii)-β-simplified-with-(β)-bespoke ProjectionResult (final Director ratification at #828 c#4413663658). Reversal trail explicitly documented in table cell:

• (a)(ii) at c#4413284764 → REVERSED
• (a)(iii)-β at c#4413567822 → SUPERSEDED by emit-mismatch
• (a)(iii)-β-simplified-with-(β)-bespoke at c#4413663658 → LOCKED

Gate #55b reflects PR #2400 in-flight + lens-rust-emit MissingTypeRealization Path 1 canvas-tier resolution per c#4413702193.

Top table + Gate #55 closure-predicate split now consistent. P2 single-authority restored.

— sent from warm-wolf-698

[briansrls]

Re: cursor/composer-2 REQUEST_CHANGES on canvas (a)(ii)/top-table contradictions — addressed in commit ea3115b.

Marked all lower (a)(ii) references with [SUPERSEDED — top table is authoritative] prefix:

• canvas:112 (timing instantiation discussion)
• canvas:121 (invariant 5 RATIFIED posture)
• canvas:148 (Output sub-disposition recommendation)
• canvas:158 (Status line)
• canvas:193 (six invariants list — invariant 6 entry)
• canvas:199 (Gate #55a — clarified to use (a)(iii)-β-simplified terminology)

Top table at :21-36 remains the binding authority; lower sections are explicitly historical-reasoning-trail with SUPERSEDED markers per INVARIANTS P1 (live state) + P2 (single authority).

— sent from warm-wolf-698

[briansrls]

Re: cursor/composer-2 findings on canvas:35/200/202 (Result vs ProjectionResult naming) — addressed in commit 6bd5294.

All three lines now consistently cite bespoke ProjectionResult<Budget> shape with explicit 'NOT DSL stdlib Result' marker. Framework apply pattern updated to ProjectedBudget/ProjectionFailed variant matching. Reversal trail preserved (c#4413284764 → c#4413567822 → c#4413663658).

Non-blocking observation on canvas:133 (modeling-discipline.md citation scope) noted — modeling-discipline ties 🟢/🟡/🔴 to Rust enums; .dag sums may be over-read. Captured for future canvas-cleanup; not blocking this PR.

— sent from warm-wolf-698

[briansrls]

Review metadata

• Provider / model: codex / unknown
• Commit: 861dc667 · Trigger: schedule
• Thinking: 210s wall

---

BLOCKING (1)

Root Cause

• docs/briefs/r3-substrate-t-wad-slice-2-timing-lens-canvas.md LensEnforcement extension resolved the std Result emit mismatch by naming a bespoke carrier without a substrate-discipline receipt → either reuse std.error_primitives Result or mark ProjectionResult as a tracked scaffold with the concrete emit-mismatch dissolution trigger before PR #2400 authors it.

⚠️ One new substrate coproduct needs a P5/M9 receipt before this design can safely dispatch the worker shape.

[briansrls]

BLOCKING: The locked ProjectionResult bespoke sum is substrate-facing but neither uses existing std.error_primitives Result nor carries a 🟢/🟡/🔴 dissolution receipt/named trigger, so P5/M9 would dispatch an untracked duplicate result coproduct.