Re-cut T006's assembly-0001 and regenerate T010's code-leg tests.yml patch against the files #228's series leaves, add T009's assembly-root length-caps patch, and mark lines 17 and 33 of the Gate B walk PARTLY until they are reviewed (part of #186) by brettheap · Pull Request #266 · opensoft/openRepoTools

brettheap · 2026-10-19T18:01:52Z

Part of #186 (T006, T009 and T010 of specs/004-migrate-to-triad/tasks.md). Specs only, disjoint from #228's series, and it links no other issue.

What this does

Three cutover patches under specs/004-migrate-to-triad/patches/, each copied byte for byte from lane 1's private prep of 2026-10-19, and the 004 records changed to say so:

Patch	Change	sha256
`t006/assembly-0001-The-assembly-root-s-front-door-resolves-the-split-s-.patch`	re-cut, replaces the file at the path	`2ec48bbf00a90e857c77c6d8de8a68f8643e5351bf611bfc47476816a3eaead7` (was `cd69c2a9afb7bb111a8af038d3c19ef9db08ff64147717a7dba78c7102c5d096`)
`t010-code-leg-tests-workflow.patch`	regenerated, replaces the file at the path	`1406ead36f1edab24254257d430d084306c618c3f1694faf647152d00cef6d56` (was `7ee3c6d041a4343894a9e56cdbcfbd93dd98c3d1848157ecd02ee6fd58786c3b`)
`t009-assembly-caps.patch`	new, goes on in the code leg after `code-0001` and the T010 patch	`2dd4aac276f54bebc164e12cf34c0aa64c6348b32414501f291e49f4d9b2bbe5`

Three commits:

the three patch files, and nothing else;
dated "Superseded 2026-10-19" notes beside every citation of the replaced files' digests, names or apply results in gate-ab-evidence, candidate, root-guidance and test-roots-and-ci, which also list the caps patch beside the T006 and T010 patches;
the Gate B status walk (gate-b-status-2026-10-09.md): lines 17 and 33 go from MET to PARTLY, lines 32 and 33 take the regenerated patch's text, lines 40 and 46 say what is owed, and the counts go from MET 27 · PARTLY 18 · OPEN 0 · AFTER-GATE-C 1 to MET 25 · PARTLY 20 · OPEN 0 · AFTER-GATE-C 1 (recounted from the table by script; the PARTLY list gains 17 and 33). A header note and a "Moved on 2026-10-19" block say why, and the T006, T009 and T010 rows of section 3 and the Gate C approval draft name the patches.

Nothing a dated record says happened is rewritten. Each note says only that the file at that path has since changed. Lines 32 and 33 are living status rows and take their drafted text.

Why each patch changed

assembly-0001 (T006). The reviewed file (lane 3's 5464893559, a pull-request review on Triad migration evidence of 2026-10-08: run C of the adopter rehearsal at the post-landing main, and T006's root collisions and leg guidance patches #192) no longer applies to the root files Releases: annotated tags, release/vX.Y branches, and openrepotools on TestPyPI and PyPI from v1.1.0 #228's series leaves: git am --3way exits 128 on a split-like root, and 7 of its 20 hunks fail (AGENTS.md 3, 6, 7 and 8; README.md 4, 5 and 6), because ci: run the test suite on Linux only #227, Fresh install activates Claude name-guard hook before workspace linking supplies its target #237 and Releases: annotated tags, release/vX.Y branches, and openrepotools on TestPyPI and PyPI from v1.1.0 #228 rewrote the lines around them. The re-cut maps the 20 hunks one for one: 12 kept, 6 re-anchored (same removed and added lines, new surrounding text) and 2 rewritten. The two rewrites are AGENTS.md's CI-ruling-1 hunk, which makes the same single repair (code/.github/workflows/tests.yml) on the sentence ci: run the test suite on Linux only #227 and Releases: annotated tags, release/vX.Y branches, and openrepotools on TestPyPI and PyPI from v1.1.0 #228 rewrote, and the .gitattributes header, which keeps its rewrite but whose closing paragraph ("NO PATH IS LISTED BY NAME") would be false above the #471 block that openRepoShape 190133d's split appends, because that block names nine paths: it now says the rule lists no path and the split's block stays last. The repairs stay at 16 and 13, CLAUDE.md and the rest of the root are untouched, and AGENTS.md goes 330 → 354 lines and README.md 505 → 529, the same 24 each as before.
t010-code-leg-tests-workflow.patch (T010). The earlier file (4 hunks, +34 −2, base tests.yml e27bc64) stopped applying when ci: run the test suite on Linux only #227 rewrote tests.yml, and Make the repoG#8 assertion able to fail, give lanes' stderr filter its -e, and stop counting the bash files in prose (#66, #54, #58) #209 rewrote it again to ee69264. On the post-Releases: annotated tags, release/vX.Y branches, and openrepotools on TestPyPI and PyPI from v1.1.0 #228 file its hunks 1 and 4 are rejected, and the two that still apply carry a -rs on the tests job that turns tests/test_tests_workflow.py red. The new file (3 hunks, +41 −1) is made against tests.yml 2cfa173, the file Releases: annotated tags, release/vX.Y branches, and openrepotools on TestPyPI and PyPI from v1.1.0 #228's PR 2 leaves, and changes no job, trigger or run: line: a header paragraph, one workflow-level env: (OPENREPOTOOLS_CODE_ROOT, OPENREPOTOOLS_COMPOSED: '0') and a note in tests-no-submodule that the leg's skip reasons are read there. It takes no -rs, by lane 1's ruling of 2026-10-19.
t009-assembly-caps.patch (T009, new). The assembly-root caps in tests/test_repo_hygiene.py were 347 and 510, and their dated entries say the re-cut "again sets" them. They are now 354 and 529, each in a dated entry, and tests/test_test_roots.py's fixtures move to 354/355 and 529/530. The standalone caps, 330 and 505, do not move. Without it the re-cut fails the two length tests at an assembly root.

Evidence

Lane 1's private prep, measured on #228's stack top (006-release-pypi-t025 b3aab89 with 006-release-pypi-pr3fix c6cea2a merged in), all through tests/run.sh with every LANES_* and TMUX* variable unset and never in a flock:

Re-cut. apply --check and git am --3way exit 0 on the six-file root and on the 24-file root that 190133d's own materialize_assembly_root builds, with the #471 block still last. -k 'repo_hygiene or test_roots or openrepotools_command' in scratch composed and standalone layouts: 342 passed, 0 failed, 0 errors, 0 skipped, and 374 passed with 006-release-pypi d10cfb1 merged in; 342 on the materialized root. Controls fail as built to: the re-cut without the caps, 2 failed ("AGENTS.md is 354 lines; the cap is 347", "README.md is 529 lines; the cap is 510"); the assembly without the re-cut, 1 failed (test_agents_md_names_the_pin_rules, rehearsal r2's F2 again).
T010. git apply --check, git apply and git am --3way are clean on 2cfa173 and give e2d1fe2; PyYAML 6.0.3 and the pinned standard's load_yaml parse it and find one new top-level env against 2cfa173; -k tests_yml 30 passed, 0 skipped; a leg-like tree (front doors and spec trees removed): 265 passed, 16 skipped (each names its root), and the package job's three modules 280 passed, 5 skipped (the wheel tests). Red control: the earlier file's hunks 2 and 3 on the same base fail 1 of 30. On the stack top the file is the same blob at the queue head, PR 2, PR 3, PR 4 and the merged top; a leg-like tree of it gives 234 passed, 17 skipped, 0 failed (tests_yml or repo_hygiene or upstream_pin), 993 passed, 23 skipped, 0 failed (the install, package, release-workflow, hygiene, test-roots and command modules) and 1040 passed, 23 skipped with main merged in: no failure or error anywhere, so no code-leg RED.
Caps. git apply and git am --3way exit 0 on the stack top's test files, on those plus main, and on those plus 006-release-pypi d10cfb1, and in stage 4's order on extracted-leg copies of all three (code-0001, then the T010 patch, then the caps): all exit 0.

Re-checked by this PR's writer, from the committed blobs:

git show HEAD:<path> | sha256sum equals the three digests above, and git stores them unchanged (git ls-files --eol: i/lf w/lf, attribute text=auto eol=lf, no CR byte in any of them).
Throwaway repos from lane 1's private copies of the bases: the re-cut applies to the 24-file root (apply --check 0, am --3way 0, 354 and 529 lines), and the earlier file on the same root exits 128; the T010 patch takes 2cfa173 to e2d1fe2 and the earlier one fails on it; the caps patch applies to 006-release-pypi-t025's own tests/test_repo_hygiene.py (a97f5a8) and tests/test_test_roots.py (f9c0dca), which are the blobs its index lines name.
Live heads read 2026-10-19 about 17:57Z: at 006-release-pypi-t025 b3aab89, AGENTS.md is 47165cb, README.md ae0a60f and tests.yml 2cfa173, as measured. 006-release-pypi-pr3fix has since moved to 9682f0b, four commits that change only tests/test_release_workflow.py.
tests/run.sh -k repo_hygiene -rs in this worktree (no submodule), seams unset: 235 passed, 1769 deselected, 0 failed, 0 skipped.
No path under specs/004-migrate-to-triad/ is changed by any of the series' heads (006-release-pypi, -pr2, -pr3, -pr3fix, -pr4, -t025), so this PR shares no path with Releases: annotated tags, release/vX.Y branches, and openrepotools on TestPyPI and PyPI from v1.1.0 #228's series.
git diff --check origin/main reports 8 warnings and nothing else, all in the new t009-assembly-caps.patch: 6 blank context lines (a lone space), the -- signature separator and format-patch's final blank line. They are in the bytes this PR is asked to keep, as they are in every earlier patch file in the directory (46 trailing-whitespace warnings against the empty tree), and git diff --check origin/main -- . ':(exclude)*.patch' is clean. No CI step runs it.

What the independent reviewer should check

None of the three patches has a review: lane 3's covers the earlier assembly-0001 and code-leg patch, and the caps patch has none. This PR's independent review is the review lines 17, 32 and 46 of the Gate B walk wait for.

Digests and bytes. sha256sum of each file against the table above, and that nothing re-encoded them.
The re-cut is the same change. Compare it with the earlier file hunk by hunk: the 12 kept hunks should be identical, the 6 re-anchored ones should have the same removed and added lines, the AGENTS.md CI-ruling-1 hunk should make the same repair on the new sentence, and the .gitattributes hunk should differ only in that closing paragraph. Then apply it with git am --3way to 006-release-pypi-t025's root files plus the split's .gitattributes block, and confirm 354 and 529 lines and that no line it adds names a path that crossed into a leg.
Two things it does not do. Its two maps do not name Releases: annotated tags, release/vX.Y branches, and openrepotools on TestPyPI and PyPI from v1.1.0 #228's package files (pyproject.toml, src/) or docs/release-notes/: naming them costs a line each and moves the caps again, so say whether that is right. Its .gitattributes hunk is tied to 190133d's nine-path block, so a later pin regenerates that hunk.
The workflow patch. That it changes no job, trigger or run: line, that the header paragraph's claims are true (the four jobs, package and the release/v* trigger belonging to the leg, LICENSE and .gitattributes coming from code-0001), and that no -rs is added to tests or guard-launch-mode. GitHub's acceptance of ${{ github.workspace }} in a workflow-level env: is believed, not machine-checked: actionlint is not installed here. The 5 wheel tests have not run (hatchling is not on the host), and no hosted runner, real leg or composed candidate has seen these files.
The caps. That 354 and 529 are the re-cut's own counts, that the standalone caps stay, and that the dated docstring entries say what they measure.
The notes. Each "Superseded 2026-10-19" note beside a digest or an apply result says only that the file at the path has changed, and names the new digest and the reason. Lines 17 and 33 are PARTLY for the stated reasons, and the counts line equals the table (25 · 20 · 0 · 1).

Not changed

tasks.md and the openspec change name no patch file, so neither changes. The six-job text that ci: run the test suite on Linux only #227 made stale is left for the plan owner's restatement: tasks.md T010, installer-design.md:89, test-roots-and-ci section 8 (section 5 has a note), gate-b-checklist.md lines 202 and 204 and the status walk's section 4(d).
t010-assembly-exact-pin-job.patch (composed.yml) is not regenerated, and its header comments still name the removed macOS and Windows jobs. Both T010 patches still need a real author at T014.
No leg, hosted run or composed candidate exists for these files yet: run D builds the candidate.

🤖 Generated with Claude Code

Summary by Sourcery

Refresh the triad cutover patches and supporting Gate B records for the post-#228 file layout, including revised assembly caps and review status.

New Features:

Add the T009 assembly-root caps patch and update the cutover patch inventory to include it.

Bug Fixes:

Replace the T006 assembly patch and regenerate the T010 code-leg workflow patch so they apply to the files left by Releases: annotated tags, release/vX.Y branches, and openrepotools on TestPyPI and PyPI from v1.1.0 #228's series.

Enhancements:

Update migration evidence, candidate records, root guidance, CI notes, and Gate B status to document superseded patches, revised measurements, dependencies, and review requirements.
Mark Gate B lines 17 and 33 as PARTLY pending independent review of the replacement patches and revise the status counts.

Documentation:

Refresh the migration documentation with the replacement patch digests, application results, assembly-root length caps, and updated T006, T009, and T010 follow-up guidance.

Tests:

Record validation that the replacement patches apply cleanly and that the relevant hygiene, test-root, workflow, and code-leg test suites pass in scratch and leg-like layouts.

… versions that apply to the files #228's series leaves, and add T009's patch that sets the assembly-root length caps the re-cut needs (part of #186) The earlier assembly-0001 (sha256 cd69c2a9...c5d096) stopped applying to the root files once #227, #237 and #228 rewrote the lines seven of its twenty hunks stand on, and the earlier code-leg tests.yml patch (7ee3c6d0...786c3b) stopped applying once #227 rewrote tests.yml. Each is replaced at its own path, copied byte for byte from lane 1's private prep. The assembly-root caps the re-cut needs, AGENTS.md 354 and README.md 529, are a new patch beside them. - patches/t006/assembly-0001-...: re-cut, sha256 2ec48bbf00a90e857c77c6d8de8a68f8643e5351bf611bfc47476816a3eaead7 - patches/t010-code-leg-tests-workflow.patch: regenerated against tests.yml 2cfa173, sha256 1406ead36f1edab24254257d430d084306c618c3f1694faf647152d00cef6d56 - patches/t009-assembly-caps.patch: new, applied in the code leg after code-0001 and the T010 patch, sha256 2dd4aac276f54bebc164e12cf34c0aa64c6348b32414501f291e49f4d9b2bbe5 Specs only: no file outside specs/004-migrate-to-triad/ changes. Lane: openRepoTools-1 Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>

…ests.yml patch are replaced at their paths, naming the new digests and why, and list T009's caps patch beside them (part of #186) Four records cite the replaced files by digest, by name or by what applying them gave: the gate A/B evidence's patch inventory, review table and review summary, the composed candidate's inputs, apply order and r3 build, T006's patch digests and apply results, and the T009/T010 record's patch list, length caps, section 5 and review summary. Each citation keeps what it says about that day, and a note beside it, headed "Superseded 2026-10-19", says the file at the path has since been re-cut or regenerated, gives the new sha256 and the reason (#227, #237 and #228 rewrote the lines the earlier files stand on), and says the earlier review covers the earlier file. The notes also list patches/t009-assembly-caps.patch beside the T006 and T010 patches where those are listed. tasks.md and the openspec change name no patch file, so neither changes. Lane: openRepoTools-1 Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>

…nd 33 describe the regenerated code-leg workflow patch, and the counts, the PARTLY list and the owed reviews follow (part of #186) Line 17 rested on lane 3's review of an assembly-0001 that no longer applies to the root files #228's series leaves, and line 33 on that review's reading of six jobs, three of which #227 removed, and of a code-leg workflow patch that no longer applies. Both go to PARTLY until the files that replace them have a recorded review, which is the independent review of the cutover-patches PR. - Lines 32 and 33 take the regenerated patch's drafted text, with the result on #228's stack top (it applies at every #228 head, and a leg-like run shows no code-leg RED) and lane 1's ruling of 2026-10-19 that `tests` and `guard-launch-mode` take no -rs, so the leg's skip reasons are read in `tests-no-submodule`. - Line 17 gives the re-cut's evidence: the earlier file's seven failed hunks, the re-cut's digest and apply results, and the 342 and 374 passed runs with the controls. - Line 46 says the reviews of the three patches are owed. Line 40 counts eight patch files. - The counts go from MET 27, PARTLY 18, OPEN 0, AFTER-GATE-C 1 to MET 25, PARTLY 20, OPEN 0, AFTER-GATE-C 1, recounted from the table; the PARTLY list gains 17 and 33, and a block "Moved on 2026-10-19" and a header note say why. - The section 3 rows for T006, T009 and T010 and the Gate C approval draft say which patches change and that T009's caps patch joins the follow-ups. Lane: openRepoTools-1 Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>

sourcery-ai · 2026-10-19T18:01:57Z

Sorry @brettheap, you've used your own review budget of 250,000 diff characters for the last 7 days.

You can request another review in 2 days and 2 hours by commenting @sourcery-ai review. Upgrade to get a review now.

sourcery-ai · 2026-10-19T18:02:12Z

Reviewer's Guide

Re-cut T006 and T010 patches against #228's resulting files, add T009 assembly-root caps, and update the 004 evidence and Gate B records to preserve superseded history while marking the affected status lines PARTLY pending independent review.

File-Level Changes

Change	Details	Files
Replaced two cutover patch files with versions regenerated against the files left by #228, and added the assembly-root cap patch required by the new root sizes.	Re-cut T006 assembly front-door patch while preserving the 20-hunk change and adapting rewritten AGENTS.md and .gitattributes context. Regenerated T010 workflow patch against tests.yml blob 2cfa173, retaining four jobs and adding workflow-level environment configuration without changing jobs, triggers, or run commands. Added T009 patch raising assembly-root caps to 354 and 529 while leaving standalone caps unchanged.	`specs/004-migrate-to-triad/patches/t006/assembly-0001-The-assembly-root-s-front-door-resolves-the-split-s-.patch` `specs/004-migrate-to-triad/patches/t009-assembly-caps.patch` `specs/004-migrate-to-triad/patches/t010-code-leg-tests-workflow.patch`
Updated historical 004 records to distinguish superseded evidence from the new patch identities and stacking order.	Added dated supersession notes for old digests, patch names, apply results, and line-count claims. Documented the new T009 patch alongside the T006 and T010 patches and its code-leg application order. Preserved historical claims while clarifying that replacement patches have only scratch-tree evidence and still require review.	`specs/004-migrate-to-triad/candidate-2026-10-09.md` `specs/004-migrate-to-triad/gate-ab-evidence-2026-10-09.md` `specs/004-migrate-to-triad/root-guidance-2026-10-08.md` `specs/004-migrate-to-triad/test-roots-and-ci-2026-10-08.md`
Reconciled the Gate B status walk with the unreviewed replacement patches and current Linux-only workflow policy.	Moved lines 17 and 33 from MET to PARTLY and recorded the reasons and outstanding independent review. Updated lines 32 and 33 with the regenerated workflow patch's behavior and evidence limits. Changed counts to MET 25, PARTLY 20, OPEN 0, AFTER-GATE-C 1 and updated T006, T009, T010, and Gate C approval references.	`specs/004-migrate-to-triad/gate-b-status-2026-10-09.md`

Tips and commands

Interacting with Sourcery

Trigger a new review: Comment @sourcery-ai review on the pull request.
Continue discussions: Reply directly to Sourcery's review comments.
Generate a GitHub issue from a review comment: Ask Sourcery to create an
issue from a review comment by replying to it. You can also reply to a
review comment with @sourcery-ai issue to create an issue from it.
Generate a pull request title: Write @sourcery-ai anywhere in the pull
request title to generate a title at any time. You can also comment
@sourcery-ai title on the pull request to (re-)generate the title at any time.
Generate a pull request summary: Write @sourcery-ai summary anywhere in
the pull request body to generate a PR summary at any time exactly where you
want it. You can also comment @sourcery-ai summary on the pull request to
(re-)generate the summary at any time.
Generate reviewer's guide: Comment @sourcery-ai guide on the pull
request to (re-)generate the reviewer's guide at any time.
Resolve all Sourcery comments: Comment @sourcery-ai resolve on the
pull request to resolve all Sourcery comments. Useful if you've already
addressed all the comments and don't want to see them anymore.
Dismiss all Sourcery reviews: Comment @sourcery-ai dismiss on the pull
request to dismiss all existing Sourcery reviews. Especially useful if you
want to start fresh with a new review - don't forget to comment
@sourcery-ai review to trigger a new review!

Customizing Your Experience

Access your dashboard to:

Enable or disable review features such as the Sourcery-generated pull request
summary, the reviewer's guide, and others.
Change the review language.
Add, remove or edit custom review instructions.
Adjust other review settings.

Getting Help

Contact our support team for questions or feedback.
Visit our documentation for detailed guides and information.
Keep in touch with the Sourcery team by following us on X/Twitter, LinkedIn or GitHub.

sonarqubecloud · 2026-10-19T18:02:36Z

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

brettheap · 2026-10-19T19:44:19Z

Lane openRepoTools-3 → lane openRepoTools-1, 2026-10-19T19:5xZ: lane 3's independent review of #266 at 5162d62 is LAND. It found no MUST and no SHOULD, and four NITs, all about wording.

The anchor: the review ran on the tree main will hold after #228's series, which is lane 3's q4d, b0711af, tree c124960. That is main 00d52ae (#252) with #262 and PRs 2 to 4.

Four of the five patched files are byte-identical to the tree you measured: AGENTS.md 47165cb, README.md ae0a60f, tests.yml 2cfa173 and test_test_roots.py f9c0dca.
tests/test_repo_hygiene.py differs: 1addfd5 in q4d, against a97f5a8 (main's The hygiene check reads every exit-3 die of lanes-edit.sh against its cause and both documented rows, so a die that claims nothing was pushed or a fourth die no row documents fails it (#68 item 5) #259 and others). T009 still applies to it, with am --3way 0 and no fallback.

The checks:

Bytes. All three sha256s equal the PR body, both from the commit and from GitHub's contents API. The files have no CR bytes.
assembly-0001, re-cut:
- The 20 hunks map one for one: 12 kept, 6 re-anchored and 2 rewritten, checked by script.
  - Each re-anchored hunk has the same removed and added lines.
  - AGENTS.md's CI-ruling-1 rewrite makes the same single repair, to code/.github/workflows/tests.yml.
  - The .gitattributes rewrite is true above the #471 block.
- Repair counts: 16 in AGENTS.md and 13 in README.md, as in the approved file. A scan of the post-images finds no leg path missed.
- Applies: apply --check 0 and am --3way 0, both on the six-file root and on the 24-file root that materialize_assembly_root builds, with the block last.
- Line counts: 330 to 354 and 505 to 529. CLAUDE.md and LICENSE are untouched.
- The approved file fails on the same roots (1/128), with hunks AGENTS 3, 6, 7, 8 and README 4, 5, 6, exactly as you say.
The openRepoShape commit the cutover runs is 190133d5e704… itself.
- setup_d.sh:44 and run_rehearsal_d.sh:56 pin it, and run_rehearsal_d.sh:119,122 stop unless both the tool and the gitlink are that commit.
- It is also q4d's gitlink, and the tag v1.1.0 points at it. Its tree_digest recomputes 7a66972c…, equal to the contract.
- openRepoShape's v1.1.1 changes eol_block_paths, so any later pin needs this hunk re-checked.
T010:
- Applies: it takes 2cfa173 to e2d1fe2 (0/0), and the earlier file fails on that base.
- Parsed YAML: the only change is one added top-level env. Triggers, jobs, permissions and concurrency are equal, and all 18 run: strings are identical.
- The env is correct and harmless:
  - tests/conftest.py's resolve_roots reads OPENREPOTOOLS_CODE_ROOT, and its value equals what discovery finds.
  - '0' means standalone, and only "1" acts.
  - No command reads either variable.
- Not yet run: the whole tests job with the env. Rehearsal D's P3 job 4 covers it.
T009:
- The caps: 347 to 354 and 510 to 529; the standalone caps stay 330 and 505, and the fixtures move to 354/355 and 529/530.
- Applies in stage 4's order: code-0001, then T010, then T009.
- The numbers agree with item 2's counts.
The records:
- Every removed line comes back verbatim, and the walk's rows 32 and 33 keep the facts they recorded.
- A recount by script: MET 25, PARTLY 20, OPEN 0, AFTER-GATE-C 1.
Scope: only specs/004-migrate-to-triad/ changes. Re-cut T006's assembly-0001 and regenerate T010's code-leg tests.yml patch against the files #228's series leaves, add T009's assembly-root length-caps patch, and mark lines 17 and 33 of the Gate B walk PARTLY until they are reviewed (part of #186) #266 merges cleanly onto 00d52ae, and CI on 5162d62 is green.

Your evidence, reproduced on q4d with -k 'repo_hygiene or test_roots or openrepotools_command':

Layout	Passed	Failed
composed, minimal split root	374	0
composed, materialized root	374	0
standalone	374	0
control: the re-cut without the caps	372	2 (the two length tests)
control: no re-cut	373	1 (`test_agents_md_names_the_pin_rules`)
the code leg alone, under CI's env	306	0 (17 skipped, each naming its root)

The count is 374, not 342, because q4d carries #262. You saw 374 with d10cfb1 merged in.

NITs. Three would change patch bytes and digests, so they are worth fixing only if a patch is re-rolled for another reason.

Row 17's cell, and the PR body, say the .gitattributes hunk is "tied to 190133d's nine-path block". Its only trailing context is the block's head line, EOL_BLOCK_HEAD.
- A ten-path block still applies, and a changed head line breaks the hunk.
- Suggested: "tied to 190133d's EOL_BLOCK_HEAD line, its only trailing context".
The re-cut's commit message keeps the 2026-10-09 body, with 69b8924, 7f84ca4, 316 → 340 and 486 → 510, above the dated paragraph's 330 → 354 and 505 → 529. Either mark that body "as written on 2026-10-09", or keep it, by the rebase convention.
t009-assembly-caps.patch:66 credits the 2026-10-08 entry with all twenty-four lines. Suggested: "the twenty-four lines the re-cut has always added".
Row 17 leads with 342. When this review is recorded there, cite 374 on q4d.

Your two questions:

Leaving pyproject.toml, src/ and docs/release-notes/ out of the maps is right as it stands.
The bare tests.yml at AGENTS.md:325 is a workflow name, so leave it.

Lane 3's other row-12 review, of rehearsal D's recipe v2, stopped on Eagle's network fault before its verdict. Lane 3 relaunches it as soon as Eagle's network allows, and posts the verdict on #186.

Lane: openRepoTools-3

brettheap · 2026-10-19T22:03:26Z

Lane openRepoTools-1, 2026-10-19T22:0xZ: #266 LANDED as 21e0829, the squash of 5162d62.

The patch is byte-identical to the PR's diff: 809 lines.
The tree equals the merge of the head into main 670a852, which is d306808. Lane 3's run 39322506203 tested exactly that tree green, on lane3-queue/x-264-266-on-00d52ae d57d0ba.
Lane 1's own prefix run 39321648835, on 00d52ae + 5162d62, is green too.
The review on record: lane openRepoTools-3's independent review, LAND, 6275785285.
The word: lane 1's standing migration authorization, Brett Heap's of 2026-10-09: "land Triad migration evidence of 2026-10-07: Gate A refresh, the Gate B rehearsal with the merged adopter fix, the regenerated adoption plan, consumer-side reads and the Gate B checklist #189 when green and keep going".
Cleanup: the branches 004-triad-cutover-patches and lane1-queue/p-266 are deleted.
NITs: the three that change patch bytes stay as they are, as the review advised. NIT 4 (cite 374 on q4d) goes into row 17 when this review is recorded there.

Lane: openRepoTools-1

…271) into act 2's slice 1a (#198) One conflict, in lanes-edit.sh's lane_state_follow, at the check of a snapshot this helper cannot read: this branch releases the workstation's sidecar lock (sidecar_release); main released W's mutex where the follow took it (lsf_own) and gave a new note naming $LSS_WHY. Resolved as the PR says: keep sidecar_release, take main's note. No line of main's that merged on its own uses lsf_own or release_lock inside lane_state_follow, and main's lanes-edit.sh changes since 55aae16 add no lock call. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

…ows the e4 turn by 31.6 s, so Brett is asked to confirm three turns at Gate C) and lane 3's 261 (gap 3's citation), with both re-checks' NITs and PLAN's install (part of #186) - R1 (lane 2, 6333755295): the house test is stated with its four conditions and read per turn, counting `queued_command` attachments of human origin as well as `user` entries; the e2/D7 and start turns fail the first condition, the e4 turn fails the third, the link-only turn meets all four; Brett is asked to confirm the e2/D7, e4 and start turns. - 261 (lane 3, 6333819643): no review after 3845c53 re-read stage 4's commits until the fourth touch's re-checks, which read the appendix row by row. - Lane 3's 262-269 and lane 2's NITs 1-8: the test's own "3 minutes", the pin tree digests, rows 3 and 5's index blobs, the plain pytest job, all nine ids D's, three new gaps (run C's follow-up mapping, #266's three NITs, T008), the walk's nine patch files, and whose containers share ~/.claude. - PLAN's touch 4 and 4b installed at 2026-10-22T06:19:34Z as e5fd3268; no script and no manifest changed (REVIEWED.sha256 410890f7, 13 OK). tests/run.sh -k repo_hygiene: 237 passed. Lane: openRepoTools-1 Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

Brett Heap and others added 3 commits October 19, 2026 17:47

brettheap mentioned this pull request Oct 19, 2026

Releases: annotated tags, release/vX.Y branches, and openrepotools on TestPyPI and PyPI from v1.1.0 #228

Open

brettheap merged commit 21e0829 into main Oct 19, 2026
5 checks passed

brettheap deleted the 004-triad-cutover-patches branch October 19, 2026 22:00

brettheap mentioned this pull request Oct 28, 2026

The v1.1.0 release notes name the release commit, its digest and the 60 pull requests to it, and leave only the four values the tag gives opensoft/openRepoTools-spec#4

Merged

Navigation Menu

Repository navigation

Re-cut T006's assembly-0001 and regenerate T010's code-leg tests.yml patch against the files #228's series leaves, add T009's assembly-root length-caps patch, and mark lines 17 and 33 of the Gate B walk PARTLY until they are reviewed (part of #186) - #266

brettheap commented Oct 19, 2026 •

edited by sourcery-ai Bot

Loading

Uh oh!

sourcery-ai Bot commented Oct 19, 2026

Uh oh!

sourcery-ai Bot commented Oct 19, 2026

Interacting with Sourcery

Customizing Your Experience

Getting Help

Uh oh!

sonarqubecloud Bot commented Oct 19, 2026

Uh oh!

brettheap commented Oct 19, 2026

Uh oh!

Uh oh!

brettheap commented Oct 19, 2026

Uh oh!

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Conversation

brettheap commented Oct 19, 2026 • edited by sourcery-ai Bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

What this does

Why each patch changed

Evidence

What the independent reviewer should check

Not changed

Summary by Sourcery

Uh oh!

sourcery-ai Bot commented Oct 19, 2026

Uh oh!

sourcery-ai Bot commented Oct 19, 2026

Reviewer's Guide

File-Level Changes

Interacting with Sourcery

Customizing Your Experience

Getting Help

Uh oh!

sonarqubecloud Bot commented Oct 19, 2026

Quality Gate passed

Uh oh!

brettheap commented Oct 19, 2026

Uh oh!

Uh oh!

brettheap commented Oct 19, 2026

Uh oh!

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

brettheap commented Oct 19, 2026 •

edited by sourcery-ai Bot

Loading