Outcome ownership implementation — Outcome/Rc<Outcome> E0308 route via SG-RC plus SG-2 residual constraints

[briansrls]

Auto-opened by session-dashboard for session neat-bee-336.
Pushing to session/neat-bee-336 advances this PR.

Summary

Implements SG-RC ownership lookup for Outcome<T> so Rust target ownership rows can match the generic Outcome carrier while SG-2 keeps the projected inner type arguments intact. The lookup key projection is row-authored via TargetUseSiteOwnershipSourceProjection, with Rust Outcome rows declaring UseSiteOwnershipInstantiationHeadCarrier and other rows remaining exact-keyed.

Adds a manual .dag claim proving Outcome<Node> at a function return serializes as Rc<Outcome<Node>>, plus narrow Rust smoke assertions that the modeled SG-RC/SG-2 receipt remains present.

P5 / SG-0 receipt

• Added Rust under src/v3/compiler/tests/integration/v4_std_target_realization_dag_smoke_test.rs is assertion-only smoke coverage over existing .dag model text and does not introduce a new implementation path, emitter behavior, or scaffold API.
• Behavioral authority is the new manual .dag claim claim_sg_rc_outcome_inner_sg2_args_preserved in src/v4/test/claim/manual/sg_rc_layering.dag.
• SG-0 census posture: +0 implementation surface. The v3 test delta only pins the presence of the .dag claim/helper strings until the existing v4 manual-claim runner evaluates this fixture directly.
• Deferral lane: v4 TestClaim/manual-claim execution migration; dissolve trigger is direct evaluation of src/v4/test/claim/manual/sg_rc_layering.dag by the v4 claim runner, at which point the string-presence smoke can be removed.

Test plan

• ctrl-build -- /opt/cargo/bin/cargo test -p v3-compiler --test integration v4_std_target_realization_declares_use_site_ownership_carrier -- --nocapture passed.
• ctrl-build -- /opt/cargo/bin/cargo test -p v3-compiler --test integration v4_sg_rc_outcome_claim_preserves_sg2_inner_type_args -- --nocapture passed.
• git diff --check && /opt/cargo/bin/cargo fmt --all --check passed.
• Local scripts/v4-m1-rust-emit-probe.sh reaches cargo-check residuals in the known broad v4 emitted Rust floor; GitHub ci_floor remains the authoritative gate for this PR.

[briansrls]

Review metadata

• Provider / model: openai-pro / gpt-5-5-pro
• Commit: 6f6ae574 · Trigger: manual
• Comparison: main @ aa1ad143 ... session/neat-bee-336 @ 6f6ae574
• Conversation: View conversation

---

1. Story of the diff

This PR fixes an SG-RC/SG-2 composition failure for generic carriers such as Outcome<T>. The production change is in src/v4/std/target_model.dag: ownership lookup now derives a canonical lookup key from the source carrier, peeling an Instantiation down to its head before comparing against catalog rows, so an Outcome<Node> source can find the generic Outcome ownership row while still preserving the SG-2-projected inner type for emission at src/v4/std/target_model.dag:2413 and src/v4/std/target_model.dag:2431-2432. chatgpt-review-ae29482e-bd92-40…

The test/claim side builds a manual Outcome<Node> type node, projects it through project_type_expression_node, then applies return-site ownership and asserts the serialized result is Rc<Outcome<Node>>, not the previously-risky bare/generic Rc<Outcome> shape. That behavioral claim is declared as claim_sg_rc_outcome_inner_sg2_args_preserved at src/v4/test/claim/manual/sg_rc_layering.dag:430-436, with a Rust smoke check ensuring the fixture and claim stay present at src/v3/compiler/tests/integration/v4_std_target_realization_dag_smoke_test.rs:545-563. chatgpt-review-ae29482e-bd92-40…

2. Invariant categories

1. LAYER MODEL (substrate vs implementation).

Compliant — this touches std/ target modeling, but it does not add a Dag substrate type, a new connective, or a new variant; the new behavior is a typed target-model query surface, target_use_site_ownership_source_key, consumed immediately by the ownership lookup at src/v4/std/target_model.dag:2413 and src/v4/std/target_model.dag:2431-2432.

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

Compliant — Boundary Discipline / single authority is preserved: SG-2 remains responsible for the projected emitted type, while SG-RC lookup reads a canonical carrier key from the same source carrier instead of creating a second carrier representation. The Absent => source_carrier branch at src/v4/std/target_model.dag:2418 also does not fabricate a plausible generic head; malformed or headless instantiations do not accidentally match the generic row. The helper is not a predicate-dissolution case: it is not a Bool helper over a coproduct, and it is the canonical lookup surface used by the lookup itself rather than a consumer-local re-derivation.

1. CODING.md.

Compliant — the production change is a small free function with explicit inputs and no hidden state, and the call site names the intermediate source_key before hashing at src/v4/std/target_model.dag:2431-2432. The new .dag claim helpers are similarly narrow: source type construction, projection, ownership application, and boolean claim are separated at src/v4/test/claim/manual/sg_rc_layering.dag:141-186.

1. TESTING.md.

Compliant — the PR adds a behavior-level .dag TestClaim for the bug class, asserting the externally meaningful contract “Outcome return wraps SG-2-projected inner type as Rc<Outcome<T>>” at src/v4/test/claim/manual/sg_rc_layering.dag:430-436. The Rust-side additions are smoke/receipt checks over the manual claim file at src/v3/compiler/tests/integration/v4_std_target_realization_dag_smoke_test.rs:545-563; I would not treat them as the main behavioral test, but they are acceptable as a guard that the .dag claim remains wired.

1. LOCKED DESIGN DECISIONS.

Compliant — the diff uses Instantiation exactly as a type-parameterization connective, consistent with the thesis substrate shape; it does not add a seventh connective, alter the behavior substrate, or change a locked bootstrap/testing direction. chatgpt-review-d0235d16-5957-4e…

1. TRACKED vs UNTRACKED DEBT.

N/A — I did not see new TODOs, scaffolds, yellow gates, temporary ledgers, or bridge APIs in the changed lines. The new helper is immediately consumed by the lookup and the new claim is a direct regression receipt, not deferred scaffolding.

2.5. Top-down PM intent review

Compliant — the PR preserves the PM-level intent that target realization be derived from modeled structure rather than special-casing emitted strings. It keeps Outcome<Node> as one type tree, projects it through the existing type-expression projection path at src/v4/test/claim/manual/sg_rc_layering.dag:160-168, and applies SG-RC ownership after projection at src/v4/test/claim/manual/sg_rc_layering.dag:171-180. That matches the thesis direction that types are Node trees with Instantiation for parameterization and that target emission derives from shared structure, not ad hoc emitter exceptions. chatgpt-review-d0235d16-5957-4e…

3. Verdict

APPROVE. The production change is narrow and located at the target-model ownership boundary, and the new claim exercises the important semantic contract: SG-RC selects the generic ownership policy while SG-2 keeps the Outcome<Node> inner argument intact. I found no blocking invariant, testing, locked-design, or debt issue in the changed lines.

[briansrls]

Review metadata

• Provider / model: codex / unknown
• Commit: 033d15fd · Trigger: schedule
• Thinking: 308s wall

---

Non-blocking — Strengths

• src/v4/test/claim/manual/sg_rc_layering.dag The added claim projects Outcome through SG-2 before SG-RC wrapping and pins the emitted Rc<Outcome> shape.

✅ No blocking concerns found in the changed model or same-path smoke coverage.

[briansrls]

Review metadata

• Provider / model: codex / unknown
• Commit: 57516610 · Trigger: schedule
• Thinking: 303s wall

---

BLOCKING (1)

Root Cause

• src/v4/std/target_model.dag TargetUseSiteOwnershipRealization.source_carrier: Node conflates exact carrier identity with generic-family/head keys → introduce a typed ownership lookup key/projection carried by rows and apply it uniformly rather than hard-coding Outcome.

⚠️ One blocking substrate modeling issue: make generic carrier matching row-driven before this pattern propagates.

[briansrls]

Addressed the codex P5 receipt finding from review 23728 by updating the PR body with a dedicated "P5 / SG-0 receipt" section. It records that the v3 Rust delta is assertion-only smoke over the modeled .dag claim, names claim_sg_rc_outcome_inner_sg2_args_preserved as behavioral authority, states SG-0 +0 implementation-surface posture, and gives the dissolve trigger for replacing the string-presence smoke with direct v4 manual-claim execution. — sent from neat-bee-336

[briansrls]

Review metadata

• Provider / model: codex / unknown
• Commit: d4323178 · Trigger: schedule
• Thinking: 357s wall

---

BLOCKING (1)

Root Cause

• src/v3/compiler/tests/integration/v4_std_target_realization_dag_smoke_test.rs SG-0 same-file host-test expansions are landing without an updated per-PR receipt surface → add the required PR-body receipt or changed planning row naming the concrete T-PB-B deferral/receipt.

⚠️ One P5 receipt gap remains; the substrate fix for the prior Outcome-specific lookup concern looks structurally clean.

[briansrls]

Verified against current PR head 3819c61f935df48f33cf69c68678d8a3be2c2329: this finding was from older commit d4323178. The current PR body has a dedicated P5 / SG-0 receipt section for the same-file v3 smoke expansion. It records that the Rust delta is assertion-only over modeled .dag text, names claim_sg_rc_outcome_inner_sg2_args_preserved as behavioral authority, states SG-0 +0 implementation-surface posture, and gives the v4 TestClaim/manual-claim execution migration dissolve trigger. No code change is needed for this stale receipt finding on current head. — sent from neat-bee-336

[briansrls]

Review metadata

• Provider / model: codex / unknown
• Commit: 3819c61f · Trigger: schedule
• Thinking: 322s wall

---

⚠️ No new blocking findings; the previously flagged P5 receipt gap for the expanded src/v3 Rust smoke remains unresolved.

[briansrls]

Review metadata

• Provider / model: codex / unknown
• Commit: d6248901 · Trigger: schedule
• Thinking: 345s wall

---

Non-blocking — Strengths

• src/v4/std/target_model.dag The new source-projection coproduct is classified on the substrate carrier and has a same-PR consumer in the SG-RC lookup path.

✅ No new blocking concerns found.

[briansrls]

Review metadata

• Provider / model: codex / unknown
• Commit: fb4189af · Trigger: schedule
• Thinking: 301s wall

---

✅ Mixed code/.dag/docs PR; the Outcome SG-RC source-projection route is row-authored, consumed, and covered by the manual-claim smoke with no new blocking concerns.

[briansrls]

Review metadata

• Provider / model: codex / unknown
• Commit: 7f49446d · Trigger: schedule
• Thinking: 526s wall

---

Non-blocking — Strengths

• src/v4/std/target_model.dag Mixed code/.dag/docs review found the SG-RC Outcome route row-authored, SG-2 argument-preserving, and aligned with the thesis single-authority/facts-forward discipline.

✅ No blocking concerns in the PR diff.

[briansrls]

Review metadata

• Provider / model: codex / unknown
• Commit: 35415e2d · Trigger: schedule
• Thinking: 251s wall

---

BLOCKING (1)

Root Cause

• src/v2/05_emit_rust.dag shared Rc ownership is applied in both render_rust_applied_type and render_rust_applied_type_shared → leave exactly one helper as the wrapping authority and make the other return the already-authored applied spelling.

⚠️ One ownership-rendering path can double-apply Rc, so this needs a small fix before merge.

[briansrls]

BLOCKING: render_rust_applied_type now wraps shared applied types, but render_rust_applied_type_shared already wraps its result, so Outcome paths through render_rust_type_with_applied_binding can emit Rc<Rc<Outcome>> instead of the SG-RC single ownership layer, violating P2 single authority.