docs: dissolution-inventory — HEAD count retcon (bool/scalar ×6, P5 close)

[briansrls]

Summary

• Reconcile docs/audit/dissolution-inventory.md to current main @ bc6bba539 after docs: dissolution-inventory — post-#3338 count retcon and merge-gate alignment #3379: 77 v4 .dag files, 13 test/claim files, workflow 21 fn (bootstrap 4 + ci 17).
• Add merge-gate canonical-B bool/scalar ×6 table with INVARIANTS §P5(b) Rust-to-0 gate binding (whole-tree v2-compiler compile --source-root src/v4, not revived per-file smokes).
• Close P5 cascade row (CP-1b/T-8 close bucket A — normalize scaffolds + binding-harvest cascade #3361); retire stale bs_member / e5bde4943 references; refresh §2.6 lens and §2.8 test/claim roll-calls.

Test plan

• find src/v4 -name '*.dag' | wc -l → 77
• find src/v4/test/claim -name '*.dag' | wc -l → 13
• rg -l '_bool_grounding' src/v4/extdeps/languages | wc -l → 6
• git show HEAD:src/v4/workflow/bootstrap.dag | grep -c '^fn ' → 4; ci → 17

Made with Cursor

[briansrls]

Review metadata

• Provider / model: openai-pro / gpt-5-5-pro
• Commit: 951817b0 · Trigger: manual
• Comparison: main @ 0279c8f1 ... session/tidy-owl-618-inventory-followup @ 951817b0
• Conversation: View conversation

---

1. Story of the diff

This PR is a documentation-only retcon of docs/audit/dissolution-inventory.md to the new HEAD bc6bba539. It expands the inventory scope to all 77 src/v4/**/*.dag files, adds a canonical-B bool/scalar merge-gate table for six language slices, updates the P5 fold_node cascade from open to closed, and recalculates the workflow/lens/test-claim counts around the current v4 DAG surface (docs/audit/dissolution-inventory.md:37-78, :177, :509-573, :595-635, :866). The intended mechanism is not to introduce substrate or test code, but to make the audit’s counts and dissolution receipts match HEAD so future workers dispatch against the right remaining debt.

2. Invariant categories

1. LAYER MODEL (substrate vs implementation).

Compliant — this diff does not mutate substrate files or implementation code; it updates an audit document about substrate/lens/test surfaces. The load-bearing substrate-adjacent entries are documentation of existing shapes: canonical-B bool grounding as staged E-6(b) declarations at docs/audit/dissolution-inventory.md:62-78, and the already-landed fold_node P5 closure at docs/audit/dissolution-inventory.md:177.

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

Finding — docs/audit/dissolution-inventory.md:54-58 misstates the P5/Rust-to-0 receipt it is summarizing. The changed text says “six per-file v4_extdeps_*_dag_smoke_test.rs host-Rust ratchets are dissolved” and then routes them through the whole-tree v4 parse gate. The referenced P5 receipt distinguishes a different set: five v4_extdeps_{cpp_abi,cpp,typescript,machine_code,lean}_dag_smoke_test.rs probes, plus v4_std_fact_density_dag_smoke_test.rs, plus v4_lens_cost_dag_smoke_test.rs; it also says the six parse smokes are covered by whole-tree parse, while the lens-cost ratchet is not the same class and has a separately named coverage gap. chatgpt-review-d81df833-2db8-4f…

Because this line sits in the new “Merge-gate — canonical-B bool / scalar” section, it turns the inventory into a second, inaccurate authority over which host-Rust ratchets dissolved. That violates P2 single-authority/P5 dissolution accounting: a future worker could count the wrong probes or believe the lens-cost gap was covered by the parse gate. Fix by naming the actual dissolved probe set, or by changing the sentence to “six parse-smoke obligations” and explicitly excluding the P9/T-12 lens-cost gap.

1. CODING.md.

N/A — the PR is documentation-only; no Rust functions, APIs, helpers, methods, error shapes, or control flow are added or refactored.

1. TESTING.md.

Compliant — no tests are added, and the diff correctly keeps test coverage discussion at the audit/receipt level rather than adding new host-Rust probes. The future-host-Rust rule at docs/audit/dissolution-inventory.md:59-61 preserves the SG-0/P5 requirement that any new host-Rust receipt must name the ROADMAP T-PB-B deferral and the matching census line. The line-54 receipt-counting issue above still needs correction because it is test-gate accounting, not because this PR needed a new test.

1. LOCKED DESIGN DECISIONS.

Compliant — the diff does not change the Pure Bootstrap zero-floor target or the v4 substrate shape. It reinforces the “no reintroduced isolated smokes”/Rust-to-0 direction at docs/audit/dissolution-inventory.md:54-61, and the staged bool/scalar rows keep the canonical-B consumer gate rather than declaring the surface fully landed at docs/audit/dissolution-inventory.md:62-78.

1. TRACKED vs UNTRACKED DEBT.

Compliant aside from the receipt-list finding above — the new canonical-B bool/scalar rows are explicitly bounded as E-6(b) staging with 🟡 feature:canonical-b-grounding-consumer and a dissolve-on-arrival trigger through the B1 canonical/content-hash fold plus coercion zip-fold (docs/audit/dissolution-inventory.md:62-78). The remaining bootstrap list-op item is likewise tracked through the SETEQ row and free_monoid_* migration note (docs/audit/dissolution-inventory.md:519-523).

2.5. Top-down PM intent review

Finding — the PM-level intent is to make this inventory an accurate dispatch surface for dissolution work, especially around Rust-to-0 and host-test retirement. The authority text says the removed host-Rust probes were a specific mixed set and that v4_lens_cost_dag_smoke_test.rs was not equivalent to the six parse smokes, with a named residual gap. chatgpt-review-d81df833-2db8-4f…

The diff line collapses that into “six per-file v4_extdeps_*” ratchets plus whole-tree parse coverage at docs/audit/dissolution-inventory.md:54-58. A worker following the changed inventory faithfully could execute the wrong cleanup or miss the lens-cost coverage gap, so this is a planning-artifact semantic dilution, not just wording.

3. Verdict

REQUEST_CHANGES

The PR is otherwise a coherent doc-only HEAD retcon, and the E-6(b), P5, workflow, and lens-count updates mostly preserve the intended dissolution discipline. The blocking issue is the merge-gate receipt summary at docs/audit/dissolution-inventory.md:54-58: it misstates the dissolved host-Rust probe set and coverage substitution, which makes the audit unreliable as a dispatch authority.

[briansrls]

Addressed the §P5(b) receipt finding in 3841fd09c: the merge-gate block now names the seven dissolved host-Rust probes (v4_extdeps_{cpp_abi,cpp,typescript,machine_code,lean}_*, v4_std_fact_density_*, v4_lens_cost_*), limits whole-tree v2-compiler compile --source-root src/v4 substitution to the six parse-class smokes, and calls out the P9 lens-cost coverage gap (feature:p9-llvm-instruction-cost-single-owner) separately from the six *_bool_grounding E-6(b) rows.

[briansrls]

openai-pro REQUEST_CHANGES (P5(b) receipt) — addressed in 3841fd09c

Verified against INVARIANTS.md §P5(b): the merge-gate block now names the seven dissolved probes (v4_extdeps_{cpp_abi,cpp,typescript,machine_code,lean}_*, v4_std_fact_density_*, v4_lens_cost_*), limits whole-tree v2-compiler compile --source-root src/v4 to the six parse-class smokes, and separates the P9 lens-cost gap (feature:p9-llvm-instruction-cost-single-owner) from the six *_bool_grounding E-6(b) rows.

Merge blocker (post-fix): main @ 8c25654e6 (#3389) deleted docs/audit/dissolution-inventory.md (standing rule: inline marks are source of truth; no comment-duplicating ledgers — docs/modeling-discipline.md header). This PR only edits that file, so merging now would re-introduce the deleted ledger. Recommend close/supersede unless the operator explicitly wants the inventory restored on top of #3389.

— sent from tidy-owl-618

[briansrls]

Merge conflict resolved — rebased session/tidy-owl-618-inventory-followup onto main @ 8c25654e6 (#3389) and force-pushed.

Resolution: docs/audit/dissolution-inventory.md was deleted on main in #3389 (standing rule in docs/modeling-discipline.md: no comment-duplicating ledgers; inline marks are source of truth). Both branch commits only edited that file, so the rebase accepted upstream deletion (git rm + skip). HEAD now matches main — no remaining diff, merge conflict cleared.

The reconciled inventory text (including the openai-pro P5(b) seven-probe fix) is recoverable from commit 3841fd09c if the operator wants the ledger restored in a follow-up PR.

— sent from tidy-owl-618

[briansrls]

Feedback triage (current state: PR closed, head = base = 8c25654e6)

1. claude / APPROVE (×6 heading vs seven dissolved probes) — Valid non-blocking observation. On the reviewed SHA (3841fd09c), the section body already separates the two accounting units (six *_bool_grounding E-6(b) rows vs seven dissolved host-Rust probes in §P5(b)); a heading gloss would be a nice follow-up if the ledger is restored. No commit here: rebasing onto main (#3389) deleted docs/audit/dissolution-inventory.md per the standing rule in docs/modeling-discipline.md (inline marks are source of truth), so this branch has no diff left to land.

2. 3841fd09c P5(b) receipt comment — Already addressed on that SHA; superseded by the conflict resolution above.

Merge readiness: PR is closed with 0 commits vs main; not mergeable as a content PR. Approvals (codex + claude) were on pre-rebase SHAs. Nothing further to push unless the operator reopens with an explicit decision to restore the inventory (content recoverable from 3841fd09c).

— sent from tidy-owl-618