NON_FOLD_RESIDUE_ROSTER re-home: 125 typed FrontierRow rows on gunbc.non_fold_residue, host census reads through, registry enrolled

[briansrls]

The dropped stretch from #7089, landed: the last walled nickname-class roster moves out of the Rust seed onto a typed .dag carrier.

Carrier — dag/gunbc/non_fold_residue.dag: 125 std.roster_frontier.FrontierRow rows (123 from the const, deduped from 127 — it carried 4 duplicate entries — plus 2 new sites below). Every dated block comment from the const is preserved as typed reason/dissolve_on fields (shared class strings; per-row receipts where a dated landing receipt existed); the two prose-only receipts (the fleet_converge_cli lens-precision bound and the 2026-07-20 stale-row ruling) are kept as notes on the carrier.

Host — cli_run.rs: the const is deleted; the census scan now reads the rows through non_fold_residue_units_from_module_source (real front-end parse, fail-closed: parse error / missing data def / non-record row / missing unit / duplicate unit / empty roster all refuse loudly, with a 🟡 dissolve-on marker to the emitted-manifest supply-carrier pattern shared with the witness-exclusion reader). nfr_roster_receipt now reports all unrostered+stale findings before asserting instead of dying on the first.

Registry — gunbc.roster_registry: the visibility note's wall retires (no known unregistered residue remains) and the roster enrolls as DeclaredFrontier (registration 14).

New residue caught by the re-home receipt: src/v2/workflow/ci_floor_plan.dag::gate_in_cheap_floor_membership + ::spec_enrolls_gate landed with #7088 unrostered (per-PR selection predict-skipped the corpus-read nfr witness — the roster's own masking class). Rostered with a dated receipt, structural-equality dissolve.

Receipts by execution

• nfr_roster_receipt: unrostered=0 stale=0 live=125 (live equivalence of the .dag roster against the corpus scan)
• reader controls: follows-synthetic-authority green; duplicate-unit and missing-unit-field refuse (RED controls)
• claim_batch: always-on witnesses (non_fold_residue_no_unrostered_or_stale + both RED fixtures) 3/3, new carrier witnesses (well-formed + populated + empty-trigger RED control) 3/3, roster-registry witnesses 5/5
• dag-only closure compiles and evaluates the carrier (gunbc run --source-root dag → true)
• generated-artifact drift gate green after ROADMAP regen (authority rows ts-group-census / ts-seed-data-out marked landed)

Also repaired in-branch: the morning merge of origin/main reintroduced stale sibling-lane content through the pre-squash branch history (17 files, namespace_graft/dag.dag/ci_floor_plan among them); all restored to origin/main byte-state — the PR diff is exactly the 7 intended files.

🤖 Generated with Claude Code

[gunbai-bot]

Review 41705's two notes, addressed:

Scope creep (the ~800-line sibling-lane delta) — real, and now fixed in da9775c. Root cause: the restore commit checked the 17 foreign files out to origin/main's current head, but GitHub renders the PR against the merge base, which was older — so sibling-lane changes that landed on main in between still displayed as PR delta. Merging current origin/main advanced the merge base; the PR files-changed view now shows exactly the 7 intended files. Receipts re-run green on the merged tree (nfr_roster_receipt: unrostered=0 stale=0 live=125; fmt clean).

Row-count reconciliation — the body's arithmetic already reconciles, with one correction to the review's note: the const held 127 entries with 4 duplicated names (each appearing twice), so 127 − 4 = 123 unique, not 122; plus the 2 new #7088 sites the receipt caught = 125 rows. No change needed.

— sent from fierce-ibex-519

[cursor]

Bugbot is not enabled for your account, so this pull request was not reviewed.

Enable Bugbot in the Cursor dashboard to get automatic reviews on future PRs.

[gunbai-bot]

Review 41712's count note verified — real drift, inverted direction: the .dag body and the live receipt are 125 (123 legacy rows deduped from the const's 127, plus the 2 fresh #7088 ci_floor_plan sites the re-home receipt caught), so the title was correct and the ROADMAP prose was stale at 123 — those authority rows were written before the receipt caught the 2 extra sites. Fixed in f34eff6: roadmap_authority.dag rows now say 125 with the 123+2 breakdown, ROADMAP.md regenerated through the artifact gate.

— sent from fierce-ibex-519

[gunbai-bot]

Review 41720's finding addressed. Verified against the sites first: the nested per-variant matches in gate_in_cheap_floor_membership / spec_enrolls_gate (ci_floor_plan.dag:97-121) are hand-rolled Gate-identity tests — each inner match g arm is literally "sg == g" spelled per-variant because Gate has no derived == today. So the accurate burn-down lane IS derived equality (once Gate inhabits equality, both folds collapse to found || sg == g and every nested match deletes) rather than an owning-file fold migration — the fold structure itself is already fine. What the review correctly caught is that the shared generic trigger string didn't say any of that. Fixed: the two rows now carry a dedicated trigger (nfr_dissolve_gate_membership) naming the exact mechanism — derived equality on Gate collapses the membership fold and deletes the nested matches with the row. Witnesses re-run green (frontier well-formed 3/3; nfr_roster_receipt unrostered=0 stale=0 live=125).

— sent from fierce-ibex-519