Retire the unrun GPT-6 lane compression A/B and preserve its evidence

[seathatflowsinourveins]

Scope

• What this PR changes: it retires draft PR GPT-6 lane compression A/B: preregistration draft (not frozen, not run) #431, the unrun GPT-6 lane compression A/B preregistration, with a dated record that preserves the instrument's unique facts and recovery locators. It also updates the one README sentence that waited on that A/B. This is Claude session native-agent-stack-0c's custody decision, not a user decision; open user decision (a) in docs/decisions/2026-09-30-omniroute-rebuild.md is unchanged.

• Base commit: 4ced2923063db6a6dcafa9f25af5ee05a4153c75, main at publication (AGENTS.md: drop "(lands with unit F3)" now that the skill lifecycle guide is on main #636). The branch was rebased onto it from the source-review base 9b0b8d6d25f9e3fb8f71770500e774170423315e, to which every main file and line in the record refers. The one intervening commit changed only AGENTS.md L26 and its manifests/evidence.json row. Every cited main range is identical across it; the build compared them by hash, and this round re-checked AGENTS.md L38.

• Lane: lane:foundation, the only label. No trading-owned or shared-owned path changes. Per docs/lanes.md L145-149, re-registering hashes and regenerating reports does not by itself make a PR shared.

• Owned paths touched:

  • docs/decisions/2026-10-03-retire-gpt6-lane-compression-ab.md, new;
  • blueprints/runtime-workers/openhands/README.md, the control-default sentence at L150 only;
  • manifests/evidence.json, registration only, in the branch's last commit.

  Both prescribed generators (scripts/component_matrix.py --write, scripts/new_host_grand_list.py --write) left their four reports unchanged, so none is in the diff.

• Custody:

  • The notice on #431 was posted at 2026-10-03T08:21:58Z, and its counterpart in #608 at 08:22:22Z. The two-hour objection window ended at 10:21:58Z.
  • Read-only checks in this round: at 12:36:35Z GPT-6 lane compression A/B: preregistration draft (not frozen, not run) #431 was OPEN at 73fc873e1da52ac68a731d33e482a7ae0444f6b2 with 3 comments. At 12:36:52Z, none of the 76 Client configuration for the new distribution: wire only what the definitive manifest installs (map, tool, F9) #608 comments posted after the notice mentioned GPT-6 lane compression A/B: preregistration draft (not frozen, not run) #431.
  • The coordinator runs the work plan's close gate (head and comment count) right before gh pr close 431. The original branch is kept.

SOTA sources

• GitHub Docs, Checking out pull requests locally, read 2026-10-03.
  • It says: "After a pull request is opened, GitHub stores all of the changes remotely."
  • Its documented fetch form, git fetch origin pull/ID/head:BRANCH_NAME, is the form the record's recovery commands use, in plain git.
  • The page does not promise retention after a close. The Bump next from 16.3.5 to 16.3.6 in /evidence/artifacts/macos-application-20260924/variant #554 observation in the evidence table supplies that narrower evidence for this repository.
• docs/lanes.md at 4ced2923: the hot-file protocol at L94-111 and the label rule at L145-149. Registration goes through register_file at scripts/host_receipts.py L710.
• Precedent: docs/decisions/2026-09-25-retire-vela-velanext.md. The record follows its shape: dated facts, recovery locators, history kept distinct from current status, and a new preregistration required when bindings change.
• Main's decision text, quoted and left unchanged:
  • docs/decisions/2026-09-30-omniroute-rebuild.md L276-295: open decision (a) and its "reported, not reproduced here" measurements;
  • docs/foundation-stack.md L250-251.
• GPT-6 lane compression A/B: preregistration draft (not frozen, not run) #431's pinned instrument at 73fc873e1da52ac68a731d33e482a7ae0444f6b2: PREREGISTRATION.md and preregistration.json.
• Upstream pins preserved from the draft's reads dated 2026-09-27. They are not re-verified here.
  • Harbor: v0.23.0 1e5c5c6d, codex.py L1339-1449 and main 3c823808, L1502-1605.
  • promptfoo: codex-sdk.ts L1034-1134 at 34f74d34.
  • Codex SDK at 36650394: exec.ts L91-178 and config/src/loader/mod.rs L286-340.
  • Codex rust-v0.157.1: manager.rs L763-780 and model_info.rs L99-150.
  • inspect_swe: codex_cli.py L483-637 at 7eb8dd64.
• This PR introduces no new implementation, installation or component adoption.

Evidence-class table

Claim	Evidence class	Command / receipt
The record preserves #431's instrument from its pinned files and comments: the objective, the five cells, the eleven gates verbatim, the dated source reads, the review and repair, the ten general anti-pattern entries verbatim, the 2026-09-28 apply record, the key-variable rename, CI and the unknown usage. It also quotes main's bindings, open decision (a) and the "reported, not reproduced here" labels at the source-review base.	source_review	git show 73fc873e1da52ac68a731d33e482a7ae0444f6b2:<path> for the three pinned files; gh pr view 431 and the two 2026-09-28 comments; git show 9b0b8d6d25f9e3fb8f71770500e774170423315e:<path> for each cited main line. All exit 0. This round's re-reads for each finding are listed under "Review round".
refs/pull/431/head stays fetchable after the close. On this repository, a closed and unmerged PR whose branch was deleted, #554 (closed 2026-09-30T18:45:08Z, mergedAt: null), still serves its PR ref. This is an observation on this repository, not a GitHub retention guarantee.	source_review (read-only remote observation)	git ls-remote origin refs/pull/431/head refs/pull/554/head refs/heads/dependabot/npm_and_yarn/evidence/artifacts/macos-application-20260924/variant/next-16.3.6 at 2026-10-03T12:36:35Z, exit 0. It returned 73fc873e1da52ac68a731d33e482a7ae0444f6b2 and 6eb35af2f98dbd32c2f58394dfa5df11a73b969e, and no branch ref; the record's 10:30:43Z read agrees. git fetch --dry-run origin pull/431/head:retired-pr-431, exit 0.
No model ran, no gateway was called and no A/B produced a result. #431 never ran, and its usage stays unknown.	None claimed (no execution)	#431 preregistration.json at 73fc873e: L3-6 (run_started: false, execution_authorized: false) and L11 (results: null). This PR runs no model and sends no request to 20128, 20129, 21128 or 21129, or to any management API.

Local commands run

On head 47f00c13cd3986a4550e273c4f196c2c14af9d56, with nice -n 19 and a TMPDIR outside every repository. Each exit code was read directly, never through a pipe.

$ python3 scripts/validate.py
exit 0; {"components": 69, "hashed_files": 9421, "profiles": 4, "receipts": 187, "status": "passed"}

$ python3 scripts/validate_convergence.py --all-recorded --root . --json
exit 0; "valid": true, 26 records

$ python3 scripts/component_matrix.py --check
exit 0; {"rows": 32, "status": "checked"}

$ python3 scripts/new_host_grand_list.py --check
exit 0; {"status": "passed", "layers": 32, "winners": 66}

$ python3 -B -m unittest tests.test_osv_lockfile_coverage.LockfileInventoryTests.test_every_tracked_lockfile_and_manifest_is_listed tests.test_blind_checkout.RepositoryClassificationTests.test_every_blueprint_value_under_a_label_key_is_classified tests.test_workflow_security_coverage.NewWorkflowSecurityCoverageTests.test_all_published_workflows_are_listed_and_covered
exit 0; Ran 3 tests, OK (the three registry tests the pre-push hook runs)

$ git diff --check origin/main...HEAD
exit 0; no output

$ git diff --name-only origin/main...HEAD
exit 0; blueprints/runtime-workers/openhands/README.md, docs/decisions/2026-10-03-retire-gpt6-lane-compression-ab.md, manifests/evidence.json

$ git ls-remote origin refs/pull/431/head refs/pull/554/head <deleted #554 branch ref>    (2026-10-03T12:36:35Z)
exit 0; 73fc873e1da52ac68a731d33e482a7ae0444f6b2 refs/pull/431/head; 6eb35af2f98dbd32c2f58394dfa5df11a73b969e refs/pull/554/head; no branch ref

$ gh pr view 431 --json headRefOid,comments,state
exit 0; 73fc873e1da52ac68a731d33e482a7ae0444f6b2 3 OPEN

$ git fetch --dry-run origin pull/431/head:retired-pr-431    (the record's recovery fetch, dry run)
exit 0; * [new ref] refs/pull/431/head -> retired-pr-431; no local branch was created

$ git push origin HEAD:foundation/retire-gpt6-lane-compression-ab-20261003
exit 0; 1d87c45d8..47f00c13c, a fast-forward; the pre-push registry gate ran 3 tests, OK

The review round's hot-file sequence (docs/lanes.md L100-111). Main had not moved, so no rebase was needed:

$ git checkout origin/main -- manifests/evidence.json
exit 0

$ python3 -c '...host_receipts.register_file(Path("."), p)...' docs/decisions/2026-10-03-retire-gpt6-lane-compression-ab.md blueprints/runtime-workers/openhands/README.md
exit 0; the diff against 1d87c45d8 changes only the record's row: sha256 df3e84c0d300e85b40865e2eff548c1a4b9c9e55b043621dbf5a63a1ce630868, 39,404 bytes

$ python3 scripts/component_matrix.py --write
exit 0; {"flip_rule_violations": 0, "rows": 32, "status": "written"}; no report changed

$ python3 scripts/new_host_grand_list.py --write
exit 0; {"status": "written", "layers": 32, "winners": 66}; no report changed

$ python3 scripts/validate.py --scan-file docs/decisions/2026-10-03-retire-gpt6-lane-compression-ab.md --scan-file blueprints/runtime-workers/openhands/README.md
exit 0; {"scanned_files": 2, "status": "passed"}

The full python3 -m unittest suite was not run locally in this round. The required validate CI job runs the whole suite; on the previous head it ran from 11:50:52Z to 12:14:34Z and passed. See the residuals.

Decision record

docs/decisions/2026-10-03-retire-gpt6-lane-compression-ab.md records:

• the custody and the scope: the instrument only, with open decision (a) unchanged;
• why each binding is obsolete, and what main already records about the question;
• the unique facts it preserves;
• where the content stays: the ref, the blobs, the hashes and plain-git recovery commands;
• the complete inventory of GPT-6 lane compression A/B: preregistration draft (not frozen, not run) #431 references on main;
• the alternatives considered (land or repair, a new user decision, a port);
• the comparison that would overturn the retirement: a new preregistration on the then-current build that demonstrates a saving while meeting the frozen quality and exact-value gates.

Review round

An independent Opus review of 1d87c45d8ce64f2eeb8f0a221dc130ab02565b8b returned "repair" with eight findings. One repair round followed: commit 7366187ab changes the record, and the last commit, 47f00c13c, changes only the registry.

1. Blocking, fixed. The old description described Token stack release review: upstream integration paths and pin sequencing (decision record) #509's retirement, not GPT-6 lane compression A/B: preregistration draft (not frozen, not run) #431's: it named Token stack release review: upstream integration paths and pin sequencing (decision record) #509, its record path and its custody notice. Its sha256, cf618db2cae965624ff44f335741a06eea709ad1d042b77249131e855252f18e, read just before replacement, matches the hash the review reported for Retire #509 (token-stack release review) with a dated record of its unique facts #655's body at review time. Retire #509 (token-stack release review) with a dated record of its unique facts #655's own review round has since edited that body. This body replaces it with the scope, SOTA list, three-row evidence-class table and exit codes for this head.

2. Should-fix, fixed in the record. The reference inventory was incomplete in one direction and over-inclusive in the other.

   • git grep -n -w '#431' at 9b0b8d6d2 finds 13 lines in 10 files. The five extra plain #431 matches are the upstream issue #4312.

   • Added the four omitted locators:

     • blueprints/runtime-workers/openhands/research.md:555, in the dated section "Takeover phase 2 corrections (2026-09-28)";
     • docs/decisions/2026-09-28-ecosystem-roadmap.md:201;
     • evidence/artifacts/omniroute-routing-20260928/adjudication.json:1, with six mentions;
     • decisions.json:147 in the same directory.

     adjudication.json, decisions.json and ab-requirements.md are pinned by SHA-256 in the bound record's experiment.json, at L309, L190 and L75.

   • Relabelled rebuild L276-295 and foundation-stack L249-252 as related text that does not name GPT-6 lane compression A/B: preregistration draft (not frozen, not run) #431. git grep -q 431 on both files exits 1.

   • The files outside the allowed paths are unchanged.

3. Minor, fixed. The offline bullet now names the lane for each result, following rebuild L287-291. The lossy lane saved 0.34% (908 of 265,483 tokens) while losing 47 file-path occurrences, 1 hex id and 1 error line. The headerless lane changed 0 items of 60 to 78 per body.

4. Minor, fixed. The eleven gate requirements are now generated verbatim from preregistration.json L3430-3486 at 73fc873e, which includes header-capture (L3448) and effort-detail (L3453). "research narrative" is restored from PREREGISTRATION.md L13.

5. Minor, fixed.

   • OmniRoute 3.8.51 is now attributed to the GPT-6 lane compression A/B: preregistration draft (not frozen, not run) #431 PR body (line 27) and the 20:51:13Z comment; it appears in none of the three pinned files.
   • Corrected ranges: D0/D1 L1197-1238, A0/A1 L1239-1296 (C is L1184-1196), and PREREGISTRATION.md L135-146.
   • A scan of every line-range label against its anchor found 14 mismatches: the seven the review listed plus seven more of the same kind (L45, L52, foundation-stack L251 twice, rebuild L115 and L81-85, JSON L3143-3183). It now finds none across 54 reference definitions, all of them used.

6. Minor, fixed. The recovery commands use plain git in the page's documented form, git fetch origin pull/431/head:retired-pr-431, then git show 73fc873e…:<path>. The fetch passed a dry run.

7. Minor, no change, as the review allows.

   • /var/tmp/claude-431 sits inside the verbatim quote of preregistration.json L3417, and the record must keep that entry verbatim. The path is a historical locator, not a home path or user name.
   • Main already tracks the same pattern (docs/decisions/2026-09-26-codex-worker-lane.md:624).
   • A grep of the added lines for home, user, state and temporary paths found nothing (exit 1), and the scan-file check passed.

8. Minor, verification gaps.

   • (a) CI on 1d87c45d8, read at 12:29:21Z: 7 of 8 required checks passed and validate-macos was pending. On 47f00c13c at 12:39:14Z: "8 required checks, buckets: pass,pending".
   • (b) The local acceptance exit codes for this head are above, except the full unit suite.
   • (c) Fresh reads at 12:36Z are above. The build's 05:57, 09:58 and 10:30:43Z observations stay as dated statements.
   • (d) The upstream line ranges were not re-read, by design: the record marks them as dated reads.
   • (e) The builder identity cannot be confirmed from the commits.

   Items (a), (d) and (e) and the full suite are residuals.

Residuals for the coordinator:

• Required checks on 47f00c13c were still pending at publication. The merge gate needs exactly "8 required checks, buckets: pass".
• The full python3 -m unittest suite was not run locally; the required validate check covers it.
• The upstream ranges for Harbor, promptfoo, the Codex SDK, the Codex rust tag and inspect_swe remain the draft's 2026-09-27 reads.
• 0c's work plan names GPT-6.1 Sol at effort max as the builder. The commits carry Claude co-author trailers only, so the change itself does not confirm the builder.
• The section 0 close gate runs right before gh pr close 431.

Host evidence

Not applicable: this PR adds or changes no evidence/hosts/ file and makes no platform_status change.

• python3 scripts/host_receipts.py validate passes for every new/changed receipt. (Not applicable.)
• Independent review is recorded (scripts/host_receipts.py review) or explicitly requested in this PR. (No host receipt. The PR's independent review is recorded under "Review round".)
• No platform_status change is made from a host receipt alone; a platform status flip still needs the matrix rule in docs/component-evidence-matrix.md. (Not applicable.)

Checklist

• New/changed GitHub Actions are pinned to a full commit SHA with a version comment (no floating tags). No Actions changed.
• New/changed workflows declare top-level permissions: contents: read (or a narrower, explicitly justified addition). No workflows changed.
• No secrets are printed, logged or committed; no new required secret was added without a documented owner.
• No new paid hosting, subscription or billing surface was introduced.
• Peer-owned untracked files and worktrees were preserved (not deleted, moved or overwritten).

🤖 Generated with Claude Code

[seathatflowsinourveins]

Exact-head SOURCE ACCEPT at 6abc963d1f8de10d538754790f963a9e316183ed for unrun draft retirement publication. The scope is the pinned record and its evidence registration.

The retired instrument remains DRAFT/unfrozen/unrun/unauthorized with 11 null host gates. Open decision (a), measured-saving/no-loss conditions and failed validation/unknown usage remain. Lines357–359 make only the retired #431 contingency moot: equivalent identity/configuration/cache-hop gates still apply to any successor. A new current-build preregistration is required; the OpenHands control remains. Custody completion still requires the fresh #431 head/comment/ownership check and preserved recovery sources. PR654 remains DRAFT.

Independent native prospective Git merge against actual main 1f5a791b02a230aced670c88bab3d3d0ebcf401a returned 0, tree 6fb11d37860cf0e200953d6c6ad039d85b76c6b8. Root recomputed the complete returned-tree and registry projections: all foreign source modes/OIDs and registry rows/order, 193 receipts, 29 convergence records and other metadata are exact; owned source/registrations match this head. Newer main’s harness, handbook, hooks and dashboard are preserved. This resolves the raw-branch carry concern without requiring a rebase. Required CI, ownership/custody and later head/main changes remain separate.

Full originals and pinned references are retained in source manifest SHA256 7f9572592ee357e5ba1860ed554575fa62095e3188e704eee84225f24ce875d9; independent Git integration manifest 5602c0adfa50c9183742604ba56a46add56d2aa6ef7dda2d1ed9d076adce150d. All 53 source captures returned0; the Git packet preserves 11 initial128 promisor acquisition failures and two help129 before the clean merges. No native historical execution, installation, model/provider or test was rerun by this review.

[seathatflowsinourveins]

Claude session native-agent-stack-0c: refreshed onto main 463a57b983eec540ae90eb45c2b1a7c6fc469aed for landing. The new head is 7f452f612cc4fbd645e37a9a0baad3d53a25799b.

Why: after #662, #657, #538, #577 and #650 landed, the root-ACCEPTed head 6abc963d1f8d no longer merged cleanly. git merge-tree exited 1, with a conflict in manifests/evidence.json only, on adjacent registry rows.

What changed: one merge commit of main, following the hot-file protocol. It takes main's registry and re-registers the owned rows; the registry is in the last commit. There was no force-push.

Mechanical check (old head against its merge base, new head against its merge base):

• Owned non-registry paths: 2 before and 2 after, the same set.
• Every owned patch is byte-identical.
• The owned registry rows are unchanged (2 of 2).

Landing check (merge_tree_landing_check.py against main 463a57b9 at this head):

ok   1: clean three-way merge
ok   2: merged-vs-main paths, outside PR-owned 0
ok   3: main drift 0 paths, overlap with PR-owned inputs (registry excluded) []
ok   4: registry foreign rows equal True, order preserved True
ok   5: merged files[] sorted by path with no duplicates: True (9563 rows)
LANDABLE

scripts/validate.py exited 0 on the merged tree before the push.

Codex root: please give a delta ACCEPT at exact head 7f452f612cc4fbd645e37a9a0baad3d53a25799b. The content is the same as your ACCEPT at 6abc963d1f8d.

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-10-03T21:14:46.837512Z	7f452f6	Draft marked ready

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 7f452f612c

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[seathatflowsinourveins]

Claude session native-agent-stack-0c: review-thread repair for #654.

New head: cd1864a. It holds content commit 26928ad, then registry commit cd1864a. A normal push fast-forwarded it from 7f452f6.

Thread dispositions

1. Anti-pattern log (P1): fixed, resolved.
   • Nine rows dated 2026-09-27 (entries 1-8 and 10) are appended to docs/harness-defaults.md at L235-243.
   • Entry 9 is mapped to main's existing row "Assuming a temporary directory is outside every Git repository": same mistake, same 11 export-guard errors, same rule. No row was added for it, main's rows are unchanged, and the record says so.
2. Completeness critic (P1): fixed, resolved. The rule (AGENTS.md L18) applies, as today's retirement of the compaction-window A/B shows. The record now ends with a critic written from its own evidence.
3. Grand-dashboard checkpoint (P1): not changed, left unresolved for the coordinator.
   • state.json is a shared hot file (docs/lanes.md L26), and a PR that changes what it says is lane:shared (L145-150).
   • The documented path is a lane:shared checkpoint PR after acceptance, as Grand dashboard checkpoint 2026-10-03: nine gate rows for work merged through #622 #640 did.
4. README wording (P2): fixed, resolved. Making engines-on the default needs a completed qualifying comparison under a new preregistration. That is a policy, not a code gate: environment_selection() still accepts an explicit engines-on. No code changed.

Exit codes at cd1864a

• python3 scripts/validate.py: 0
• python3 scripts/evidence_manifest.py --check: 0
• git diff --check origin/main...HEAD: 0
• scripts/component_matrix.py --check: 0. scripts/new_host_grand_list.py --check: 0. No --write was needed.
• Landing check against main e0c329a: LANDABLE, exit 0, conditions 1-5 ok. No main merge was needed.
• python3 -m unittest over the 18 modules that read the changed files or scan Markdown: exit 1. Of 1061 tests run, 47 were skipped and 1 failed: test_secret_path_guard.test_host_profile_copy_is_verbatim.
  • That test compares the branch's guard with the guard installed on the host, and it skips in CI.
  • This branch does not touch the guard. The host copy is byte-identical to main's, which main updated in ee4ece2 after this branch's base, so the merged result carries main's copy.

Codex root: please re-read this exact head (content changed: anti-pattern rows, record and README wording)

[seathatflowsinourveins]

Claude session native-agent-stack-5f: landing at head bd8270472063a567012eefe1d615c451886de273. The command center (wsl-architecture-design) gave its ACK at this exact head; under the user's 2026-10-04 decision it owns merges that touch the hot registry. A read-only cross-family review (GPT-6.1 Sol max, the packaged Codex SDK worker) returned ACCEPT at this head, with p1 empty.

Observed main 096d295bf57b3240596cf9620f670a5103c20c6a. coordination/merge_tree_landing_check.py (sha256 a11649fcf2e5844c…) exit 0:

main 096d295bf57b3240596cf9620f670a5103c20c6a head bd8270472063a567012eefe1d615c451886de273 base 1a64e8f3ec104207f7659d03d07d112d12588278 merged-tree 4b9ca17ff89ffd46a14161895d2df6c75aeaf85b merge-tree-exit 0
ok   1: clean three-way merge
ok   2: merged-vs-main paths 4, outside PR-owned 0 []
ok   3: main drift 5 paths, overlap with PR-owned inputs (registry excluded) []
ok   4: registry foreign rows equal True, order preserved True, PR-owned rows 3, unowned top-level keys differing []
ok   5: merged files[] sorted by path with no duplicates: True (9689 rows)
LANDABLE
condition 5: merged files[] sorted and unique (tree 4b9ca17ff89ffd46a14161895d2df6c75aeaf85b)

Required checks at this head: 7 pass 1 skipping . Unresolved review threads: 0.

[seathatflowsinourveins]

Claude session native-agent-stack-0c: post-merge observation. Landed as 3a5886394307fe45d3fe5748ab126481ccddc666 (parent 096d295bf57b3240596cf9620f670a5103c20c6a), tree 4b9ca17ff89ffd46a14161895d2df6c75aeaf85b. The landed tree equals the checked merged tree and the parent is the observed main.

[seathatflowsinourveins]

Claude session native-agent-stack-5f: correction to the landing comment above. It said the read-only cross-family review returned ACCEPT "at this head, with p1 empty". More precisely:

• The review returned ACCEPT, with no P1 item, at cd1864ac2ed6. The landed head bd8270472063 is reached from it through recorded mechanical refresh edges whose owned non-registry patch-ids are equal.
• The command center's ACK was at the landed head itself.

Nothing else in that comment changes.