# Gate record — releases#331 W4 (update sources, #339), PR #265 @ `1e981dcb`

Gater: doyle, 2026-09-24 15:29Z, HFENDULEAM.
Lane: todlando. Branch `feat/339-update-sources`. Base `b96b88d2` (hertz's #264, IR-149 thin-lane
docs-drift, landed earlier this session).

## Subject verified before any green was read

- Head `1e981dcbfb2cd9498b3c5484b4980beb1548633e`, and run `36018501396`'s `headSha` is that exact
  sha — the run measures the gated tree, not a lookalike.
- `git merge-base --is-ancestor origin/main 1e981dcb` true against `origin/main = b96b88d2`, and the
  merge-base IS `b96b88d2` — a true fast-forward.
- Exactly two commits ahead: `9a77dc68` feat(update): update-source ladder, peers-first composite,
  source verbs / `1e981dcb` test(update): source_ladder_e2e, heavy at birth. 13 files, +1931/-49.
- Rebased twice onto `b96b88d2` during the lane; the third base move was deliberately avoided by
  holding hertz's docs lane rather than landing it first.

## Conditions

| # | condition | verdict | evidence |
|---|---|---|---|
| 1 | CI green on the gated sha | HOLDS | run `36018501396` on `1e981dcb`, **6/6**: traceability, changes, lint, unit Linux (kitsubito), unit Windows (hfenduleam), docs-drift |
| 2 | traceable-reqs clean | HOLDS | traceability job success at the gated sha; todlando's local `traceable-reqs check` EXIT=0, 0 findings; `xtask check` OK locally |
| 3 | nextest arm | HOLDS — doyle ruled | 1077/1077, `nextest_exit=0`. The brainproc ready-breadcrumb red seen earlier was PROVEN a single-process runner artifact (PASS at 146/1077 under process-per-test), not W4 |
| 4 | diff items outside the product surface ruled | HOLDS — doyle ruled | see below |
| 5 | ff-only land, tested sha == merged sha | HOLDS | pushed `1e981dcb:refs/heads/main` from `b96b88d2`; PR #265 state MERGED, `mergeCommit.oid == 1e981dcb`; `origin/main` re-read post-op = `1e981dcb` |

## Diff items I ruled (both outside todlando's nominal product lane)

- **`golden.yml` HEAVY entry stays with todlando** — HEAVY-at-birth: a new test's partition entry is
  its birth certificate and rides the commit that creates it. Routing it to hertz opens a window
  where a cross-node e2e exists un-partitioned in Phase A on the single self-hosted slot. Added:
  +1 Phase B for `source_ladder_e2e`, +1 Phase A for `source_verbs_e2e`. hertz was told it was there
  so he would not meet a `golden.yml` edit for the first time DURING a golden run; he agreed on the
  reasoning.
- **`docs-site/src/cli/reference.md` +98 lines accepted** ON the predicate that they are GENERATED
  bytes — the `xtask check` docs-drift gate is exactly the instrument that proves that, and this PR
  is the FIRST product-code PR through that gate. It ran green, so the gate now has a positive
  production sample and not only its own self-tests.

## Notes carried forward

- `docs-drift` green here is the first product-code sample of the IR-149 gate; the thin-lane union
  gating it landed at `b96b88d2` immediately before.
- hertz's docs lane (`docs/ir149-150-records` @ `83a12d6a`) was HELD through this gate solely to
  avoid a third base move for this lane. Released at land time.
- W5 (#338 apply side) carries the two H2 unmeasured items — bundle-less set must read as
  "nothing to apply", and the `./` tar member prefix — pinned as required unit rows in the W5 REQ
  seed, `.spt/preserved/331/dispatch-todlando-w5.md`.