Repository navigation
release: refresh the rc.1 candidate pins to the heads of main on 10 Nov 2026 - #22
BotshareAI wants to merge 3 commits into
Conversation
…ov 2026 Nine of the eleven OpenAMRobot pins move from the 2026-10-19 heads to the heads of main on 2026-11-10 (interfaces, comm, manipulation, platform-sw, upper-body-sw, both hardware repos, ui, docs). Both firmware pins and the two upstream driver pins are unchanged. The candidate now contains the combined model (upper-body-sw #14), map saving (platform-sw #115) and the base masses (platform-sw #126). Still a candidate snapshot, not rc.1 acceptance. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01DZnPERXdK4VNyx7trxz3QR Signed-off-by: Alex Reznichenko <info@botshare.ai>
…rm pin tools/check_upstream_pins.py reads upstream_pins.yaml from the imported workspace and fails when openamr_platform_sw differs from the manifest's platform-sw pin. Runs in manifest-validation after the exact export. Fails on the current candidate (2ca7392 records 131e500, manifest pins 713574f) until upper-body-sw #48 merges and its merge commit is pinned. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01DZnPERXdK4VNyx7trxz3QR Signed-off-by: Alex Reznichenko <info@botshare.ai>
|
Status at head
Generated by Claude Code |
panthera-momagdii
left a comment
There was a problem hiding this comment.
Reviewed the cross-pin check and tests at af8309d. The check is correct and fails on this candidate for the expected reason: upperbody main 2ca7392 records platform 131e500, while the candidate manifest pins platform 713574f. The exact-head manifest workflow fails at that cross-pin check; repository quality passes. Keep this PR Draft. After #48 is accepted and merged, replace the upperbody pin with the actual merge commit, then rerun the full import/export and manifest checks. #48 now has A6/A7 green at a9d3349, but its CLA and Brinex adoption gates remain open. No approval or waiver of the cross-pin failure.
KARTHIKEYAN124
left a comment
There was a problem hiding this comment.
Reviewed af8309d. Hosted manifest validation correctly rejects upperbody’s platform pin 131e500 against the manifest’s 713574f; the Jazzy driver build passes.
Before marking ready:
- Merge the accepted upperbody #48 fix, select its resulting commit and rerun cross-pin validation.
- Link full Gate A acceptance and owner approval of the refreshed snapshot.
- Keep interfaces repinning separate: restore its existing pin here or obtain an explicit owner decision for this change.
- Coordinate a release-repository PR using identical shared SHAs and regenerated checksums. Preserve release #12’s agreed 19 October candidate scope.
Hardening suggestion: use the existing strict duplicate-key loader for upstream_pins.yaml.
I inspected the diff and hosted failure log; no independent import, ROS build or hardware test was run.
panthera-momagdii
left a comment
There was a problem hiding this comment.
REQUEST_CHANGES at exact head af8309d294983a6eef92a1ffa3d7236ecb88df7a for one new validator defect. I rechecked the candidate manifest, cross-pin checker, strict manifest loader, previous review threads and exact-head hosted jobs. The manifest-validation failure is the expected platform/upperbody mismatch (131e500 vs 713574f); keep this PR Draft until upperbody #48 is accepted and the manifest pin is refreshed to its actual merge commit. I also verified that interfaces #29 is merged at 14912e18ecb6473ebd7ef4a42c4bd1870b2023f6, exactly the interfaces pin in this candidate. Under the stated heads-of-main snapshot that pin is consistent; if the release scope intends to retain the old interfaces pin, record that owner decision explicitly.
New P2: the cross-pin checker uses yaml.safe_load for upstream_pins.yaml, so duplicate mapping keys are silently accepted (last value wins). I reproduced a file with two openamr_platform_sw entries: the first contains a wrong 40-character SHA, the second contains the expected SHA; safe_load returns the latter and cross_pin_errors() reports no error. This lets a malformed upstream pin file conceal a conflicting pin. Reuse the existing duplicate-rejecting StrictLoader (or equivalent) for this file and add a regression proving duplicate pins keys fail closed. The current upstream_pins.yaml itself has no duplicate; this is a validator robustness defect, not a claim that the present candidate's mismatch is hidden.
| continue | ||
| path = Path(workspace) / holder / rel | ||
| if not path.is_file(): | ||
| errors.append(f'{holder}: {rel} not found in the workspace (import the release first)') |
There was a problem hiding this comment.
P2: yaml.safe_load silently keeps the last duplicate key. I reproduced two openamr_platform_sw entries (first wrong SHA, second expected SHA); this checker then returns no error. Use the existing duplicate-rejecting StrictLoader and add a duplicate-key regression so an upstream file cannot hide a conflicting pin.
KARTHIKEYAN124
left a comment
There was a problem hiding this comment.
Acknowledged. The cross-pin checker needs the duplicate-rejecting StrictLoader and regression tests for duplicate mapping keys. After that fix, rerun CI. The PR will remain draft until upperbody #48 is accepted and its actual merge commit is pinned, resolving the platform/upperbody mismatch. The interfaces-pin scope will be recorded explicitly.
|
Governance clarification at unchanged head |
Signed-off-by: Karthikeyan <mahikarthik10@gmail.com>
|
Parking note for rc.1: #23 carries the v2.0.0-rc.1 pins (the 20 Nov reviewed main of each repository, from the release runbook). This PR's 10 Nov pins are not the rc.1 pins, and its cross-pin checker would fail rc.1 while upperbody-sw main pins platform-sw 8e7a454. Please do not merge it for rc.1. Its owner can close it as superseded for the pins, or rebase the checker alone after rc.1. Nothing here is closed or deleted; review history stays. Generated by Claude Code |
Refreshes the v2.0.0-rc.1 candidate pins from the 19 Oct heads to the heads of main on 10 Nov 2026. It is still a candidate snapshot, not rc.1 acceptance; the final selection needs owner review.
Precondition: repository openAMRobot/openamrobot-manifest, branch
release/rc1-pins-2026-11-10, parent3bf169e(main), expected outcome a draft PR changingreleases/v2.0.0-rc.1.reposandSTATE.md. Second commit (parent51588ef): adds the cross-pin checktools/check_upstream_pins.py, its test and one CI step.What changes
Nine of the eleven OpenAMRobot pins move to the heads of main on 10 Nov 2026. The development manifest, paths and URLs are unchanged.
8a7a5b1713574ff57e54c2ca7392c5558ad0ca798b87ab8a914912e11c5529273fa694095e6c7be76d306c32fba8ed2b066349ff2e76012695c614086c706bUnchanged: both firmware pins (
f96014dandba1a53f, mains unchanged since 12 Oct) and the two upstream driver pins (OA-05). The header comment reads "Candidate snapshot: 2026-11-10 (heads of main)". STATE.md gets a dated section.Cross-pin check (new):
tools/check_upstream_pins.pyruns in manifest-validation on the imported rc.1 workspace and fails when openamr-upperbody-sw'supstream_pins.yamlnames a different platform-sw commit than this manifest. It fails on this candidate on purpose: upper-body2ca7392records platform-sw131e500, the manifest pins713574f. This PR stays draft until upper-body-sw #48 merges and its merge commit replaces the upper-body pin.Tests
3bf169e, headaf8309d.python3 -m pytest -q tests: 45 passed (9 subtests), including 3 new intest_upstream_pins.py. Withoutyamllintinstalled, 9 tests fail on main as well; with it installed, all pass on main and on this head.python3 tools/check_release_manifest.py releases/v2.0.0-rc.1.repos --development openamrobot.repos: PASS development coverage and sources; PASS immutable refs, 13 repositories.python3 tools/check_manifest.py releases/v2.0.0-rc.1.repos(live): every one of the 13 SHAs fetched and verified.vcs import /tmp/ws < releases/v2.0.0-rc.1.repos, thenvcs export --exact /tmp/ws, thencheck_release_manifest.py ... --exact-export: PASS, the exact export matches all release pins.yamllint -d '{extends: relaxed, rules: {key-duplicates: enable, line-length: disable}}' releases/v2.0.0-rc.1.repos openamrobot.repos: exit 0. The only warnings are about colon spacing in the unchanged development manifest.colcon listat the workspace root: 20 packages. The duplicateopenamr_ui_msgs(ui and interfaces) is still there, and so are the three UI packages underui/, outsidesrc/. Both are known (OA-08, the UI layout) and unchanged by this PR.python3 tools/check_upstream_pins.py releases/v2.0.0-rc.1.repos --workspace <ws>with upper-body at2ca7392:FAIL src/openamr-upperbody-sw pins openamr_platform_sw 131e500fc4e5, the manifest pins src/openamr-platform-sw 713574fd2a5a, exit 1 (expected).cross_pin_errorsdisabled,test_upperbody_built_against_other_platform_failsfails (1 failed, 2 passed). The pin refresh itself is data; the checks above run on the pins.Not verified
vcs validatecrashes on vcstool 0.3.0 (version_typeUnboundLocalError, issue Manifest check: vcstool 0.3.0 validate fails on a commit that is a branch tip #12). The full-SHA live check above replaces it, as the repo already does.openamr_ui_msgs(OA-08).a9d3349: A6 and A7 passed; code owner review changes requested onceb6b90). The upper-body pin2ca7392was built against platform-sw131e500, while this manifest pins713574fwith the new base masses. The cross-pin check reports this.Integration Gate
check_manifest.py,check_release_manifest.py, manifest-validation.yml).Dependencies
None changed in this repository. vcstool, yamllint and colcon were installed locally to run the checks.
AI disclosure
Claude Code (Claude Opus 5.5) read the heads of main, edited the pins and STATE.md, ran the checks and wrote this description. No safety-path logic is touched.
Generated by Claude Code
Incremental validator fix ? 13 November 2026
Precondition: openAMRobot/openamrobot-manifest, branch release/rc1-pins-2026-11-10, parent af8309d; new head 72753a0. Addresses review 5862248493 without changing any pins.
Reused the existing duplicate-rejecting StrictLoader for upstream_pins.yaml. YAML failures now produce cross-pin validation errors. Regression controls reject repeated platform keys and repeated top-level pins mappings even when the final value matches the manifest. Existing matching/mismatching-pin controls remain.
Tests:
python -m pytest -q tests/test_upstream_pins.pybefore the implementation fix reproduced both duplicate-key failures (2 failed subtests, 4 passed). After the fix,python -m pytest -q tests/test_upstream_pins.py tests/test_release_manifest.py: 12 passed, 9 subtests passed.git diff --check: passed. Full localpython -m pytest -q testson Windows: 38 passed, 11 subtests passed, 8 failed because existing temporary-fixture writes produce CRLF rejected by yamllint. This is not a full-suite pass; hosted Linux results must be checked separately.Not verified: no new source import, ROS build, simulation or hardware test was performed for this validator-only change. The known platform/upperbody mismatch remains and must continue to fail CI until an accepted upperbody merge commit is selected. Keep this PR draft. Gate A acceptance, coordinated release refresh and owner confirmation of interfaces-pin scope remain separate requirements; release #12 retains its 19 October snapshot.
Dependencies: none changed in the repository. AI disclosure: Codex implemented the strict-loader fix, regression tests and STATE note and ran the local checks above. No acceptance decision or release publication is implied.
Hosted verification at 72753a0: run https://github.com/openAMRobot/openamrobot-manifest/actions/runs/43483262217 ran 46 regression tests successfully (OK). The workflow then failed as expected on upperbody platform pin 131e500fc4e5 versus manifest platform pin 713574fd2a5a. Repository quality, DCO and CLA passed. The separate Jazzy driver run was still in progress when recorded; no success is claimed for it here.