docs(roadmap): §6a MethodContract dissolution triggers + v3 lens inventory

[briansrls]

R2 Release — r2-release-6a-follow-through-worker.md (lens slice + ROADMAP)

Site inventory (brief §1)

Field	src/v3/lenses/cost.dag	src/v3/lenses/complexity.dag
size_effect from *_templates()	none	none
cost_shape from *_templates()	none	none
callback_element_position from *_templates()	none	none

§6a:173 demo path method_contract_cost_shape on MethodContract remains the live consumer surface in cost.dag.

Bulk migration (brief §2)

N/A for current v3 lens sources — no lens-local *_templates() reads to retire. Acceptance for the cost.dag / complexity.dag slice is satisfied by null inventory (R2 Release accepted on inbox).

ROADMAP / dissolution triggers (brief §3)

• New ### §6a \MethodContract` per-field dissolution triggerssubsection under## Tracked debts`.
• Three named rows: size_effect, cost_shape, callback_element_position — each cites §6a:175 + upstream lane (HOC row TBD per brief).
• Algebra asymmetry bullet updated with v3 lens slice audit sentence.

Commit: 115ad625b (tip of session/proud-gull-114 vs main is ROADMAP-only at file level; earlier #968 work is already on main).

Brief acceptance checklist

• Site inventory (above; null for v3 lens slice)
• cost.dag + complexity.dag: no *_templates() reads of the three fields (vacuously met)
• cargo test --workspace --exclude v2-compiler-tests — run in CI / dev (author env lacks cargo)
• cargo clippy --all-targets -- -D warnings — run in CI / dev
• DB-8 fixed-point — CI / dev
• ROADMAP dissolution-trigger rows (this PR)

Scope note

Live (algebra_id, method_id) MethodContract producer + call-site consumer wiring stays under Release Goal 5 / program lanes (E-P / compiler), not this PR.

Made-with: Cursor

Made with Cursor

[briansrls]

Review metadata

• Provider / model: cursor / composer-2
• Commit: 115ad625 · Trigger: schedule
• Thinking: 220s wall

---

Verdict: APPROVE — The diff tightens ROADMAP §6a with explicit per-field dissolution conditions and a dated v3 lens read-surface note, resolves Tier 2 §5 in the synthesis brief with an audit-backed receipt, and lands a focused integration gate plus SG-0 census wiring. Nothing here contradicts INVARIANTS, modeling discipline, CODING, or TESTING in a way that needs a blocking fix; the ROADMAP claim that src/v3/lenses/cost.dag / complexity.dag do not touch *_templates() matches the tree (only method_contract_cost_shape on MethodContract in cost.dag), and lower.rs still has exactly two dag.push_node(Behavior::Loop( sites, matching the new test’s expectation.

Exploratory (optional): The construction-closure test in r2_b5_loop_construction_closure_test.rs (e.g. the needle / assert_eq!(count, 2) block) is deliberately brittle to refactors that split or rename the push_node(Behavior::Loop( call pattern; that is an accepted ratchet tradeoff already called out in the updated synthesis brief, not a rubric violation.

[briansrls]

Review metadata

• Provider / model: codex / unknown
• Commit: 115ad625 · Trigger: schedule
• Thinking: 90s wall

---

⚠️ Unable to review: the required source-of-truth command gh pr diff 990 --repo gunb-ai/gunbc failed because this environment cannot resolve api.github.com.

[briansrls]

proud-gull-114 — dashboard queue item (association + checkout note)

This item is routing/context, not a code review thread: it only records that this session is tied to #990 and that the PR branch might need a local checkout before edits.

Verification: gh pr view 990 shows MERGED into main from session/proud-gull-114 (tip included 115ad625 / ROADMAP §6a + prior v3 integration fixes in the same branch history). There is no remaining open finding on #990 for a follow-up commit from this queue line.

Local worktree: In this agent environment the linked git worktree (gitdir: …/gunbc/.git/worktrees/proud-gull-114) is not usable (missing primary repo), so I cannot re-checkout the branch here; any new work should start from an up-to-date main clone or a repaired worktree.

No code change required for this item.