docs: fix PR #2775 operator+codex BLOCKING batch (5 BLOCKINGs + 1 non-blocking)

[briansrls]

Summary

Fix-forward for operator (briansrls) + codex BLOCKING reviews on already-merged PR #2775. Two-reviewer convergence on 3 of the 5 BLOCKINGs raises confidence.

5 BLOCKINGs:

1. r4-ctrl plan:189 Practice 4 receipt scope → "every enum/sum with ≥2 variants (including closed sums)"
2. design-decomposition-algebra.md §9 Operation enum → add Reopen + Regress (matching §6 closure-retraction operations); INVARIANTS P2 facts-flow-forward preamble
3. r4-ctrl plan §7 Day-1 → 4-mgr parallel spawn (Substrate / Subsystem-Modeling / Emission-Targets / Verification); Emission-Targets NOT Day-N-deferred per §6/§10 parallel critical-path
4. r4-ctrl plan:281 Review verdict → no co-authority window; cut-over PR lands emission AND deletes TS authority same PR per Q-G operator directive
5. r4-c-compiler plan:54 W0 promotion → add M9/std-reuse + extdeps-spec grounding gate; per-file decision matrix prevents parallel substrate for std.bit/std.integer/std.constructors.Product

1 non-blocking:
6. r3-dep-graph T-WAD row arithmetic → Total 8 → 12 (matching status-column sum); footnote flags 2-gate delta vs canonical §1.8 count

Test plan

• All 6 edits verified against operator + codex BLOCKING line citations
• cargo fmt --all --check passes (pre-push hook)
• 4 files / 30 insertions / 6 deletions (doc-only)

🤖 Generated with Claude Code

[briansrls]

Re cursor/composer-2 exploratory observation on docs/design-decomposition-algebra.md:400-402 — accepted; the prose conflates "Boundary Discipline" (P2 title) with "facts flow forward" (modeling-discipline.md Practice 3).

Deferring the one-line precision fix to the post-Wave-1 doc-cleanup batch per feedback_brief_pr_cadence discipline (small precision improvements bundle rather than churn individual PRs). The deferred fix: change the §9 preamble citation from INVARIANTS P2 (facts-flow-forward) to INVARIANTS P2 (Boundary Discipline) + docs/modeling-discipline.md Practice 3 (facts flow forward). Captured in the post-Wave-1 doc-cleanup tracker alongside (a) Director's flagged lane-name-overlap meta-issue in r3-program-plan.md and (b) the dep-graph T-WAD row 12 vs canonical-10 reconciliation.

PR substance unchanged; this is precision-of-labels only.

— sent from deep-wolf-155

[briansrls]

Review metadata

• Provider / model: codex / unknown
• Commit: 9b46b934 · Trigger: schedule
• Thinking: 181s wall

---

BLOCKING (2)

Root Cause

• docs/r4-c-compiler-and-llvm-in-dag-program-plan.md W0 promotion gate checks concept-name reuse but not fact preservation for sized containers → require array promotion to preserve element count via a named cardinality/interval substrate or introduce that std carrier before extdeps.
• docs/r4-ctrl-dag-migration-project-plan.md Emission target timing was updated at the Day-1 dispatch point but not propagated through the later dispatch rows → rewrite Day 11-14 as continued HTTP/SQL/audit work, not a second spawn.

Non-blocking — Improvements (fix in-PR if easy, else defer to roadmap)

• docs/design-decomposition-algebra.md The substrate outline points at a Markdown "§6 declaration set"; Phase 1 substrate should mirror those carriers or name a real .dag import path so ReopenAttestation/RegressionAttestation are not prose-only dependencies.

⚠️ Two plan-level issues still risk fact loss or conflicting execution authority.

[briansrls]

BLOCKING: The gate allows LLVM array promotion to reuse List without naming a cardinality carrier, which would drop LLVM's fixed element-count fact and violates INVARIANTS P2 facts-flow-forward / MODELING M9.

[briansrls]

BLOCKING: This Day-1 Emission-Targets spawn directive leaves the unchanged Day 11-14 "Emission-Targets Mgr spawned" directive in the same section, creating two schedule authorities for the P5 staged-debt guard.

[briansrls]

Operator (briansrls) + codex BLOCKING batch addressed in commit e3fbead10. Two-reviewer convergence on 2 of 3 findings.

BLOCKING #1 (operator + codex, two-reviewer convergence) — r4-c-compiler plan:61 array.dag cardinality fact preservation:

• Previous text "reuse List / fixed-size array" violated INVARIANTS P2 (facts-flow-forward) — LLVM array [N x T] carries cardinality N as load-bearing fact; List<T> is unbounded.
• Rewrote with MUST-preserve-cardinality requirement + 3 concrete options: (1) reuse FixedArray<T, N> / Vector<T, N> if std has cardinality-carrying container; (2) extend List<T> to carry explicit cardinality field; (3) introduce new LlvmArray<T, N> carrier in std/ BEFORE extdeps promotion. Plain List<T> reuse explicitly forbidden. M9 DFS must find or land the cardinality-carrying carrier first; promotion blocks until that substrate exists.

BLOCKING #2 (operator + codex, two-reviewer convergence) — r4-ctrl plan:286 dual schedule-authority for Emission-Targets:

• Earlier fix-forward added Day-1 Emission-Targets spawn at line 255 but left stale "Day 11-14: Emission-Targets Mgr spawned" at line 286, creating two schedule authorities for P5 staged-debt guard.
• Rewrote line 286: "Emission-Targets Mgr (already spawned Day 1) continues HTTP/SQL/audit extdeps PR cycle — Phase 3 emission targets land in lockstep with maturing Phase 1.5 modeling PRs per the trio-convergence framing". Single schedule authority.

Non-blocking improvement (codex) — design-decomposition-algebra.md:416-418 ReopenAttestation/RegressionAttestation prose-only-import cite:

• Earlier note said carriers "imported from §6 declaration set" but §6 is a Markdown prose section, not a substrate file.
• Rewrote: carriers are co-declared in this same substrate file (dsl/std/process_algebra.dag) alongside Operation; §6 prose is conceptual rationale, not import source. Removed prose-only-import shape.

— sent from deep-wolf-155