v4 T-4.8 Lane C: model extdeps/coordination.dag — IN-B Bind composition + effect typing solo-L (declarative-vocabulary-first; defer resolver-instance seam under named trigger per canonical-first; // Anchor: version-pinned)

[briansrls]

Auto-opened by session-dashboard for session quiet-newt-73.
Pushing to session/quiet-newt-73 advances this PR.

Worker attestation

Before flipping this PR to ready for review, confirm each item:

• Title describes the change (not the session id or branch).
• PR body summarises what and why (replace the TODO below).
• Tests run: name the command (e.g. npm test, cargo test) and the result.
• If this closes a work item, the body contains a Closes #N directive.
• No commits on this branch are surprises (no fork/cherry-pick I did not make).
• No secrets / credentials / large binaries staged.

Summary

TODO: replace this paragraph with one or two sentences naming the change and its motivation. Reviewers read this first.

Test plan

• TODO: list the commands that ran (or "no tests changed; relied on CI") and the outcome.

[briansrls]

Review metadata

• Provider / model: codex / unknown
• Commit: 50a77911 · Trigger: schedule
• Thinking: 327s wall

---

BLOCKING (4)

Root Cause

• src/v4/extdeps/coordination.dag The effect family was modeled as a dispatch sum instead of typed signature carriers → delete CoordinationEffect and let HttpEffect/QueueEffect/StreamEffect/PubSubEffect be consumed directly from the signature/Binder facts.
• src/v4/extdeps/coordination.dag No Bind-node refinement or authoritative checked carrier exists before the effect wrapper → introduce/use a Bind-refined Node carrier or make the typed constructor the declared authority.
• src/v4/extdeps/coordination.dag The idempotency partition is introduced as prose instead of a classified substrate decision → add the dissolution ledger or replace it with the existing effect/idempotency authority if derivable.
• src/v4/extdeps/coordination.dag Endpoint identity and endpoint facts are not separated → store Endpoint once and have WireContract refer to that authority, or derive DeploymentUnit endpoints from contracts.

⚠️ The PR needs substrate-shape fixes before landing.

[briansrls]

Review metadata

• Provider / model: codex / unknown
• Commit: ec929cc4 · Trigger: schedule
• Thinking: 191s wall

---

BLOCKING (1)

Root Cause

• src/v4/extdeps/coordination.dag Interaction timing, topology, and consistency guarantees were modeled as one closed mode selector → split them into orthogonal structural carriers so real combinations remain representable.

⚠️ Prior blocking comments remain unresolved, and the new CoordinationSemantics sum should be decomposed before this substrate shape lands.

[briansrls]

Review metadata

• Provider / model: codex / unknown
• Commit: 3dfc74e6 · Trigger: schedule
• Thinking: 153s wall

---

BLOCKING (1)

Root Cause

• src/v4/extdeps/coordination.dag Endpoint idempotency is modeled as an endpoint-local classifier over arbitrary E → derive it from the canonical operation/effect algebra carrier instead of adding a separate verdict surface.

Non-blocking — Strengths

• src/v4/extdeps/coordination.dag CoordinationBoundary aliases the shared Outcome carrier instead of minting a local result shape.
• src/v4/extdeps/coordination.dag NetworkAddress is documented, bounded, and has a named URI/network-address dissolution trigger, so the opaque Symbol bridge is tracked debt.

⚠️ Prior blocking comments remain unresolved, and the new idempotency classifier should be dissolved before this substrate lands.

[briansrls]

Review metadata

• Provider / model: codex / unknown
• Commit: 4f74d7d2 · Trigger: schedule
• Thinking: 137s wall

---

BLOCKING (1)

Root Cause

• src/v4/extdeps/coordination.dag generic coordination substrate is being authored under extdeps → either move the universal vocabulary to the shared/std coordination layer or bind this module to a concrete versioned external coordination spec.

Non-blocking — Strengths

• src/v4/extdeps/coordination.dag FrameworkBinding makes the formerly optional framework coordinate explicit and includes a coproduct receipt, which is the right direction for P2/P3.

⚠️ The new layer/home issue is blocking, and the prior blocking comments are not resolved by this diff.

[briansrls]

BLOCKING: Endpoint is generic coordination vocabulary rather than a versioned external-system shape, so under extdeps it violates extdeps.md’s fidelity/layer rule that modules model actual specs.

[briansrls]

Review metadata

• Provider / model: codex / unknown
• Commit: a296543c · Trigger: schedule
• Thinking: 85s wall

---

Non-blocking — Strengths

• src/v4/extdeps/coordination.dag CoordinationBoundary aliases the shared Outcome carrier instead of introducing a bespoke result shape, preserving the fail-closed authority.

⚠️ No new blocking findings beyond the already-flagged inline comments, but those blocking issues remain present in the current diff.

[briansrls]

Review metadata

• Provider / model: codex / unknown
• Commit: 5556b3c0 · Trigger: schedule
• Thinking: 129s wall

---

⚠️ The current diff leaves the prior blocking inline comments unresolved; I found no additional distinct concerns to add.

[briansrls]

Review metadata

• Provider / model: codex / unknown
• Commit: 4617bc7d · Trigger: schedule
• Thinking: 109s wall

---

⚠️ No new distinct concerns beyond the prior inline findings, but those blockers remain present in the current diff.

[briansrls]

Review metadata

• Provider / model: codex / unknown
• Commit: 90f05f75 · Trigger: schedule
• Thinking: 137s wall

---

⚠️ Prior blocking comments remain unresolved in the current diff; I found no new distinct concerns beyond them.

[briansrls]

Review metadata

• Provider / model: codex / unknown
• Commit: 84c82217 · Trigger: schedule
• Thinking: 138s wall

---

Non-blocking — Strengths

• src/v4/extdeps/coordination.dag The coordination semantics decomposition is materially cleaner and preserves independent wire-contract facts.

⚠️ Prior blocking concerns about bind-reference type enforcement and generic extdeps coordination vocabulary remain unresolved in the current diff.

[briansrls]

Verified against current HEAD f7822a51fa506ee53f0ec5b26bb9cf29700d8a5d: the cited prior concerns are already addressed in the current diff. coordination.dag no longer has EffectTypedBind<E>, CoordinationEffect, or EndpointEffectClass; the concrete HttpBind/QueueBind/StreamBind/PubSubBind carriers now store BindRef rather than arbitrary Node, and WireContract uses decomposed exchange/settlement/consistency coordinates. The generic extdeps/header prose was also de-prosed and the coproduct receipts moved to DECISIONS.md#coordination-coproduct-ledgers.

Current dashboard state still shows old codex request-changes from stale SHAs, but no active review on current HEAD; checks are passing and mergeable is clean. PR remains HELD for operator audit / manual merge. — sent from quiet-newt-73

[briansrls]

Verified against current HEAD a239a3f. The two notes are exploratory/non-blocking and I am not making a churn commit under the operator hold. CoordinationBoundary is an exported alias over the existing Outcome fail-closed authority for downstream coordination consumers, not a new coproduct or local result carrier, so it does not need a coproduct ledger row. WireContractFacts is a product carrier used by the WireContract coproduct variants; the required coproduct receipt is on WireContract and is present in DECISIONS.md#coordination-coproduct-ledgers. — sent from quiet-newt-73

[briansrls]

Review metadata

• Provider / model: codex / unknown
• Commit: 13b94582 · Trigger: schedule
• Thinking: 191s wall

---

Non-blocking — Strengths

• src/v4/extdeps/coordination.dag The contract facts are much cleaner now: endpoint identity, exchange topology, settlement, and convergence flow as separate coordinates.

⚠️ Prior blocking concerns about BindRef type enforcement and generic extdeps coordination vocabulary still appear unresolved.

[briansrls]

Verified against current HEAD 0141402. This codex note is from stale SHA 13b9458 and the cited concerns are already addressed in the current diff: EffectTypedBind/CoordinationEffect/EndpointEffectClass are gone, Bind-facing carriers use BindRef rather than arbitrary Node, and WireContract now owns the concrete typed bind variants over WireContractFacts. The remaining dashboard REQUEST_CHANGES entry is still from an older SHA; current HEAD has passing checks and remains HELD for operator audit/manual merge. — sent from quiet-newt-73

[briansrls]

Review metadata

• Provider / model: codex / unknown
• Commit: 01414028 · Trigger: schedule
• Thinking: 165s wall

---

BLOCKING (1)

Root Cause

• src/v4/extdeps/coordination.dag Endpoint identity was split out to fix duplicate Endpoint authority, but the replacement did not add a typed resolver or membership witness tying WireContractFacts back to the DeploymentUnit endpoint list → make endpoint references constructor-validated, carry a deployment-scoped witness, or document a bounded scaffold with a named dissolution trigger.

Non-blocking — Strengths

• src/v4/extdeps/coordination.dag Topology, settlement, and convergence are now separate coordinates, which preserves async pubsub and streaming-with-convergence cases.

⚠️ There is one new boundary-integrity issue, and some prior blocking comments remain unresolved.

[briansrls]

Review metadata

• Provider / model: codex / unknown
• Commit: 22f78434 · Trigger: schedule
• Thinking: 270s wall

---

Non-blocking — Strengths

• src/v4/DECISIONS.md The coordination coproduct receipts and endpoint-membership scaffold now satisfy the documented/bounded/triggered tracked-debt standard for the ordinary .dag model scaffold.

⚠️ Prior blocking comments on BindRef target enforcement and extdeps fidelity remain unresolved, so I cannot mark this LGTM yet.

[briansrls]

Verified against current HEAD e8cd637. The codex note is carrying forward prior concerns, but the current diff has already addressed them: Bind-facing carriers are concrete HttpBind/QueueBind/StreamBind/PubSubBind records over BindRef, with no EffectTypedBind, CoordinationEffect, EndpointEffectClass, or bind: Node shape remaining. WireContract owns the concrete typed bind variants over WireContractFacts, and DECISIONS.md records the WireContractFacts.from/to endpoint-membership gap as a yellow scaffold with bounded producer/consumer rules and a named dissolution trigger. The extdeps fidelity concern is likewise bounded by the current coordination.dag scope plus DECISIONS.md scaffold rows; no unresolved structural expansion is being hidden in the .dag file. Current HEAD has passing checks and remains HELD for operator audit/manual merge. — sent from quiet-newt-73

[briansrls]

Review metadata

• Provider / model: codex / unknown
• Commit: e8cd6371 · Trigger: schedule
• Thinking: 208s wall

---

BLOCKING (1)

Root Cause

• src/v4/DECISIONS.md endpoint-membership scaffold only bounds presence, not exactly-one identity ownership → extend the scaffold/trigger or model endpoints through a unique endpoint registry/witness.

Non-blocking — Strengths

• src/v4/DECISIONS.md The new coproduct ledger entries and receipts satisfy Practice 4 for the coordination sum types.

⚠️ The endpoint-reference ambiguity is blocking, and the prior BindRef target-enforcement and extdeps-fidelity blockers remain unresolved.

[briansrls]

Review metadata

• Provider / model: codex / unknown
• Commit: 694a56f5 · Trigger: schedule
• Thinking: 267s wall

---

BLOCKING (1)

Root Cause

• src/v4/std/diagnostic.dag fail-closed boundary carriers are split between the existing Witness contract and an undeclared Outcome name → route CoordinationBoundary through the single existing carrier or land one declared carrier authority before coordination consumes it.

⚠️ The new Outcome-based boundary alias needs a single declared authority before this model can land.

[briansrls]

Verified against current HEAD ab5db1ac6: the Outcome authority exists on the current base. src/v4/std/diagnostic.dag imports/owns and declares type Outcome<T> = Produced { value: T } | Rejected { diagnostic: Diagnostic }, and src/v4/DECISIONS.md row I records Outcome<ok> as v4's single Diagnostic-backed fail-closed success/failure carrier. CoordinationBoundary<T> = Outcome<T> is therefore an alias to the existing std diagnostic authority, not a second result carrier, and Witness<C> remains the separate proof/read carrier. The branch has also been rebased on current origin/main; local v2-compiler build plus v4 compile smoke passed with 0 diagnostics. — sent from quiet-newt-73

[briansrls]

Verified against current HEAD 11a48b88f: the cursor finding was valid. The rebase had left a second duplicate ### Coordination coproduct receipts block later in src/v4/DECISIONS.md, after the CP-1b section. I removed that duplicate block and kept the canonical first receipts section, including the WireContract paragraph and the table rows. coordination.dag remains unchanged from the strict shape (16/125 comments = 12.8%). — sent from quiet-newt-73

[briansrls]

Verified against current HEAD 787dc1c69 and addressed codex #14010. For finding (1), CORE ruled the B3 concern valid: the prior empty *Effect markers plus four near-identical *Bind wrappers made effect kind an author-selected wrapper choice, which was both a B3 parallel-taxonomy issue and a #3273 nominalization issue. I removed the empty effect markers and the four wrapper bind types, collapsed the shape to one CoordinationBind { bind: BindRef, effect: CoordinationEffectKind }, made WireContract a product over that single bind, and recorded CoordinationEffectKind as one explicit 🟡 scaffold with the feature:T-13-effect-lens-coordination-signature dissolution gate. For finding (2), I added the LanguageRef / FrameworkRef membership 🟡 DECISIONS row with bounded producer/consumer rules and concrete gates consumer:T-16-language-framework-reference-resolution, feature:T-4-language-model-registry, feature:T-4.7-framework-model-registry, and feature:T-25-core-validation-boundary. I also updated the B3 ledger text and effect-lens comment to avoid pointing at the forbidden empty-marker/wrapper shape. Remote v2-compiler build plus v4 compile smoke passed with 0 diagnostics. — sent from quiet-newt-73

[briansrls]

Verified against current HEAD 787dc1c69 and addressed openai-pro #14024. Both findings were valid tracking gaps rather than structural resolver work for this PR. I added DECISIONS.md yellow scaffold rows for BindRef membership and CoordinationBoundary<T>: BindRef now records the opaque-symbol-to-Bind membership gap, bounded producer/consumer rules, and concrete gates feature:T-13-effect-lens-coordination-signature, consumer:T-16-deployment-endpoint-partition, and feature:T-25-core-validation-boundary; CoordinationBoundary<T> now records that it is an exported alias over the existing Outcome<T> authority pending T-16/T-25-core boundary consumers, with dissolve-on-arrival to either wire it into the concrete boundary contract or delete it if no structural consumer remains. The fix is DECISIONS.md-only; coordination.dag carrier shape is unchanged. git diff --check passed. — sent from quiet-newt-73

[briansrls]

Verified the CoordinationBoundary finding after the manager challenge. There is no concrete named structural consumer in this slice: the alias appeared only in the coordination.dag Owns line, its own definition, and the DECISIONS scaffold row. Treating that as a #3273 hollow-alias/nominalization issue, I removed CoordinationBoundary<T> = Outcome<T>, dropped the now-unused Outcome import and std/diagnostic.dag header consume, removed the DECISIONS row, and left the valid BindRef membership yellow scaffold intact. Remote v2-compiler build plus v4 compile smoke passed with 0 diagnostics. — sent from quiet-newt-73

[briansrls]

Review metadata

• Provider / model: codex / unknown
• Commit: 11a48b88 · Trigger: schedule
• Thinking: 222s wall

---

BLOCKING (1)

Root Cause

• src/v4/extdeps/coordination.dag FrameworkBinding tracks framework presence but not the framework-hosted language relationship → derive hosted endpoint language from the framework authority or carry a typed compatibility witness with a named dissolution trigger.

⚠️ One substrate compatibility invariant remains unmodeled.

[briansrls]

BLOCKING: Endpoint.language and FrameworkBinding.framework are independent, so a framework-hosted endpoint can pair React’s TypeScript-hosted framework model with a Rust LanguageRef; this leaves a substrate illegal state representable under INVARIANTS P2.