Prepare v1.4.2 Board clarity release source change - #1006
Merged
Merged
Conversation
Bump version to 1.4.2, add v1.4.2 release notes/qualification docs, move current release guidance (README, quickstart, install, pypi-release runbook, etc.) forward to the v1.4.2 identity, and add tests/test_release_v142.py for the version/runbook/Board-restart identities. v1.4.0/v1.4.1 tags, release notes, qualification docs and tests remain untouched and immutable. No cloud event fields are added or changed. #951's merged local-evidence code is treated as accepted; its bounded hosted Devin canary remains pending and is not claimed here. Part of #952. Closes #952. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
…ic-release-checklist.md docs/public-release-checklist.md said "the current package-index release entrypoint is `code-mower==1.4.2`" and "the corresponding GitHub tag is `v1.4.2`" in the same section that opens by saying v1.4.2 is not yet published or qualified -- an install-facing contradiction, not merely an established convention worth preserving. The current published entrypoint is `code-mower==1.4.1` (tag `v1.4.1`); v1.4.2 is the target entrypoint after its acceptance under #952. Rewrote the bullet to say exactly that, and added tests/test_release_v142.py::PublicReleaseChecklistCandidateStatusTests plus an updated tests/test_release_hygiene.py::test_current_release_docs_record_package_index_procedure to hold the corrected wording and reject the old ambiguous phrasing. Verification (fresh venv, unittest's own exit code, no pipeline masking): - `python -m unittest tests.test_release_v142` -> 20/20 passed, exit 0. - `python -m unittest discover -s tests -p "test_release_v14*.py"` -> exit 1 overall, exactly one failure: the byte-identical, untouched historical `test_release_v141.InstalledPromptPackTests.test_literal_starter_and_explicit_config_walkthrough` (confirmed via `git diff HEAD:<path> <path>` -> no output for that file, docs/v141-release-notes.md, docs/v141-qualification.md, and docs/v140-release-runbook.md). - `python -m unittest discover -s tests -p "test_release_hygiene.py"` -> 361/361 passed, exit 0. - `ruff check .`, `scripts/privacy_scan.py`, `scripts/guard_package_workflows.py`, and `python -m code_mower.migration release-readiness --json` (status: pass) all clean on this commit. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
jeffhuber
marked this pull request as ready for review
September 17, 2026 05:48
Contributor
Author
Codex audit (merge-authority lane)Head SHA: Codex Audit: BLOCKED Summary: The release identity changes leave two existing tests incompatible with version 1.4.2, blocking CI. The workflow digest mismatch was verified directly; the full suite was not run in the read-only environment. Findings:
|
Both independent-audit findings on c3af1dd (PR #1006 comment, CI run 35187134784) are fixed narrowly, without touching or consolidating any other historical v1.4.1 file: - tests/test_release_v141.py::InstalledPromptPackTests.test_literal_starter_and_explicit_config_walkthrough built and installed the CURRENT checkout, then asserted `code_mower.__version__ == '1.4.1'`. With this PR's version bump that installs 1.4.2 and fails under ordinary `unittest discover` on every CI Python version. This historical walkthrough only qualifies a supplied v1.4.1 artifact (see docs/v141-qualification.md); the `CODE_MOWER_QUALIFICATION_WHEEL`-supplied path is untouched and still asserts `1.4.1` exactly as before. Only the no-wheel-supplied branch changed: instead of building the current (now 1.4.2) checkout and misreporting it as 1.4.1, it skips with an explicit reason. Nothing else in the file changed -- confirmed via `git diff` against c3af1dd, a four-line net change inside one method. - tests/test_lineage_producer_artifacts.py::ArtifactTests.test_default_init_emits_activation_with_standalone_pure_helper normalized only `refs/tags/v1.4.1` before comparing .github/workflows/release.yml against its frozen SHA-256 baseline, so this PR's `v1.4.2` tag bump broke the digest match. Extended the normalization to also replace `refs/tags/v1.4.2` before hashing, restoring the baseline digest while every other workflow byte is still compared exactly. docs/v141-release-notes.md, docs/v141-qualification.md, docs/v140-release-runbook.md, and tests/test_release_v142.py are confirmed unchanged from c3af1dd (no diff). No consolidation, rename, or deletion of any historical v1.4.1 file. Verification (fresh venv, unittest's own exit code, no pipeline masking): - `python -m unittest discover -s tests -p "test_release_v14*.py"` -> 27 tests, exit 0, OK (skipped=1) -- the previously-BLOCKED version-mismatch failure is gone; the one skip is the no-wheel-supplied branch above. - `python -m unittest tests.test_release_v142` -> 20/20 passed, exit 0 (identical to c3af1dd's result; file unchanged). - `python -m unittest discover -s tests -p "test_lineage_producer_artifacts.py"` -> the fixed test passes; 3 other, unrelated failures in that same file are pre-existing (wheel-build/`/tmp` symlink path issues in this sandbox, present on unmodified main) and out of scope for these two findings. - `python -m unittest discover -s tests -p "test_release_hygiene.py"` -> 361/361 passed, exit 0. - `ruff check .`, `scripts/privacy_scan.py`, `scripts/guard_package_workflows.py`, and `python -m code_mower.migration release-readiness --json` (status: pass) all clean on this commit. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Contributor
Author
Codex audit (merge-authority lane)Head SHA: Codex Audit: PASS Summary: No actionable regressions were identified. Runtime validation remains incomplete because the environment lacks PyYAML. Findings: none. |
Contributor
Author
|
Owner gate reconciliation for exact head
|
jeffhuber
added a commit
that referenced
this pull request
Sep 17, 2026
v1.4.2 is published from 55339bf under the annotated v1.4.2 tag, and release #952 is closed, but current-facing docs on main still described it as an unpublished source candidate pending #952. - State v1.4.2 as the published package-index baseline across README, install, quickstart, try-in-10-minutes, the public-release and OSS checklists, the early-adopter runbooks, sessions, github-setup, builders-grok-cursor, the PyPI runbook, and release history. README, current-state, and the rollout plan now use versioning.public_baseline_sentence() verbatim. - Convert docs/v142-release-notes.md and docs/v142-qualification.md into final records. Only proven outcomes are PASS, each bound to its evidence: the independent Codex audit of PR #1006 head 32706cf, CI run 35188065537, publish run 35189302150, the wheel/sdist digests, and both verified local Board restarts. Run 35189721623's skipped publish jobs are recorded as the intended posture. Metadata upload is marked not required; no cloud aggregate check is claimed. Both pages note that the immutable v1.4.2 tag still carries the prepublication snapshots; the tag is not touched. - Record v1.4.0/v1.4.1/v1.4.2 as shipped in the roadmap, Board complete at 55339bf, #951 as the separately pending hosted canary, and v1.5.0 Slack as the active phase. Historical PR mapping moves to past tense and merged PRs now use /pull/ URLs while issues use /issues/. - Surface Graphify as shipped optional functionality: README navigation to setup/lifecycle/queries, a ramp-up flow in graphify-setup.md, the excluded inputs and the absence of a working-tree watcher, and a dated historical banner on graphify-evaluation.md that preserves its benchmark record. - Surface the persistent Board service and its macOS/launchd boundary from README and docs navigation, fix the board-demo browser wording, and move the stranded delayed_health_failed row into the delayed-health section where its post-apply semantics belong. - Add official pipx/uv installation links and command -v preflight checks without any curl-pipe-shell, and record the install boundaries adoption feedback keeps returning to. - Sync templates/lanes/README.md with the packaged copy so both document the supported `never` token expiry. - Make README links absolute so the built PyPI long description resolves, and teach release_readiness to accept either link spelling. - Canonicalize the installed-lineage harness temp root and both sides of every provenance assertion so the documented replay is portable on macOS, where temporary directories arrive through the /var -> /private/var symlink. - Move CHANGELOG Unreleased to the top and empty it; its entries shipped in v1.4.1 and are recorded there. - Rewrite tests/test_release_v142.py away from candidate-state enforcement onto the published identity, README link behavior, packaged-template consistency, issue-vs-pull URLs, and the repaired lifecycle table. Refs #1008. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
7 tasks
jeffhuber
added a commit
that referenced
this pull request
Sep 17, 2026
…fy boundary (#1009) * Docs: reconcile published v1.4.2 guidance, roadmap, and qualification v1.4.2 is published from 55339bf under the annotated v1.4.2 tag, and release #952 is closed, but current-facing docs on main still described it as an unpublished source candidate pending #952. - State v1.4.2 as the published package-index baseline across README, install, quickstart, try-in-10-minutes, the public-release and OSS checklists, the early-adopter runbooks, sessions, github-setup, builders-grok-cursor, the PyPI runbook, and release history. README, current-state, and the rollout plan now use versioning.public_baseline_sentence() verbatim. - Convert docs/v142-release-notes.md and docs/v142-qualification.md into final records. Only proven outcomes are PASS, each bound to its evidence: the independent Codex audit of PR #1006 head 32706cf, CI run 35188065537, publish run 35189302150, the wheel/sdist digests, and both verified local Board restarts. Run 35189721623's skipped publish jobs are recorded as the intended posture. Metadata upload is marked not required; no cloud aggregate check is claimed. Both pages note that the immutable v1.4.2 tag still carries the prepublication snapshots; the tag is not touched. - Record v1.4.0/v1.4.1/v1.4.2 as shipped in the roadmap, Board complete at 55339bf, #951 as the separately pending hosted canary, and v1.5.0 Slack as the active phase. Historical PR mapping moves to past tense and merged PRs now use /pull/ URLs while issues use /issues/. - Surface Graphify as shipped optional functionality: README navigation to setup/lifecycle/queries, a ramp-up flow in graphify-setup.md, the excluded inputs and the absence of a working-tree watcher, and a dated historical banner on graphify-evaluation.md that preserves its benchmark record. - Surface the persistent Board service and its macOS/launchd boundary from README and docs navigation, fix the board-demo browser wording, and move the stranded delayed_health_failed row into the delayed-health section where its post-apply semantics belong. - Add official pipx/uv installation links and command -v preflight checks without any curl-pipe-shell, and record the install boundaries adoption feedback keeps returning to. - Sync templates/lanes/README.md with the packaged copy so both document the supported `never` token expiry. - Make README links absolute so the built PyPI long description resolves, and teach release_readiness to accept either link spelling. - Canonicalize the installed-lineage harness temp root and both sides of every provenance assertion so the documented replay is portable on macOS, where temporary directories arrive through the /var -> /private/var symlink. - Move CHANGELOG Unreleased to the top and empty it; its entries shipped in v1.4.1 and are recorded there. - Rewrite tests/test_release_v142.py away from candidate-state enforcement onto the published identity, README link behavior, packaged-template consistency, issue-vs-pull URLs, and the repaired lifecycle table. Refs #1008. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> * Docs: restore backslash continuations in the Graphify ramp-up commands Four multi-line commands in the new ramp-up flow lost their trailing line-continuation, joining each command onto one line with stray whitespace. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> * Docs: distinguish the published v1.4.2 Graphify integration from main PR #1007 merged to main at b863e63 after v1.4.2 was published, so the two are no longer the same integration. Current docs still described #1007 as an open, unmerged pull request and left its entry out of the changelog entirely. - CHANGELOG: Unreleased is no longer empty. It carries #1007's Fixed entry and states the boundary directly -- accepted on main, in no published package, intended for the next appropriate release. The v1.4.1 shipped entries and the immutable 1.4.2 section are unchanged, and the Graphify fix is not double-booked into 1.4.2. - graphify-setup.md: replace the stale "separate open pull request ... not merged and not released" paragraph with a "Published `v1.4.2` versus current `main`" section naming what merged, that the accepted 0.9.58 pin and wheel digest are untouched because this is a Code Mower fix rather than a provider upgrade, and that a generation built before the next release must be rebuilt with `context-graph refresh`. A published generation is never rewritten in place, so upgrading alone does not repair an older partial frontend generation. - current-state-and-roadmap.md: the published package carries the originally shipped integration; the merged compatibility fixes are recorded as on main awaiting the next appropriate release, in the Graphify phase and in the delivery-order note about merged-but-unpublished work. - README, context-graph-lifecycle.md and context-graph-queries.md: state where the #1007 behaviour does and does not apply, so a reader on the published package is not told the separate provider-manifest budget, `doc_ref` exclusions, or JavaScript/TypeScript test conventions are available to them. - graphify-evaluation.md: point the historical banner at that boundary section. - test_release_v142.py: drop the assertion that #1007 is absent from the changelog, which encoded the now-false claim that it is unmerged. Replace it with coverage that Unreleased carries the merged entry, that the setup and roadmap pages state the boundary in both directions including the rebuild requirement, that no current page still calls #1007 open, and that the pin stays at 0.9.58. Rebased onto b863e63 with #1007's changelog entry and documentation preserved. Refs #1008. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> * Docs: scope the v1.4.2 Graphify boundary and narrow the rebuild claim Two documentation inaccuracies found in root review, both in the post-v1.4.2 Graphify guidance. "Everything above describes that package" was false. PR #1007 added the language-extras and runtime-ownership paragraphs to "Separate acquisition environment", which sits above the boundary section, so part of "above" describes current main rather than the published v1.4.2 package. The claim is now scoped to the base setup and ramp-up -- acquisition, the separate contained offline build, and steps 1 through 7 -- and the two #1007 paragraphs carry an explicit post-v1.4.2 marker where a reader meets them, which the boundary section names. The rebuild guidance implied every generation built before the next release must be rebuilt. It is narrowed to the generations #1007's compatibility gaps actually affected: most often an older frontend generation left partial, whose oversized provider manifest was refused or whose inputs a missing language parser could not process. A generation "context-graph status --json" already reports usable needs no rebuild. Applied in docs/graphify-setup.md, README.md, and docs/current-state-and-roadmap.md; the changelog and release records carried no equivalent claim. Two tests in tests/test_release_v142.py enforce the boundary: one rejects any "everything above describes that package" spelling and requires the scoped claim plus the in-place post-v1.4.2 marker above the boundary section, the other rejects blanket rebuild-everything-built-before wording across all three pages and requires each to tie the rebuild to the partial state and exempt a generation status already reports usable. Both fail against the pre-fix text. No product code changed. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> * Require a path-segment boundary for README public-doc links _links_to_repository_doc accepted any destination whose path ended with the required relative path, so docs/OTHER_SUPPORT.md satisfied the SUPPORT.md requirement and the public-docs-linked-from-readme check could pass on a README that never links the real file. Accept a destination only when its path is exactly the relative path or ends with "/" + the relative path. Both spellings the check exists for still pass: the relative form in the repository README and the absolute GitHub blob URL the packaged long description needs. Query strings are now dropped alongside fragments, since neither changes which file the destination resolves to. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> * Resolve README public-doc links instead of matching their suffix The path-segment boundary added in 58976cc still let two destinations satisfy the public-docs-linked-from-readme check without addressing the required file: a nested relative path that does not exist in this tree, such as docs/SUPPORT.md for SUPPORT.md, and any unrelated absolute URL whose path ends in /SUPPORT.md, including another owner's or another repository's GitHub URL. Resolve the destination to the one repository-relative path it lands on, then require that path to equal the required document. A relative destination is normalized against the repository root, where README.md sits, so docs/SUPPORT.md resolves to itself and no longer matches. An absolute destination resolves only when it is this repository's own GitHub URL -- github.com/codemower-ai/code-mower/{blob,raw}/<ref>/<path> -- reduced to the path under its ref; every other host, owner, and repository resolves to nothing. Query strings and fragments are still dropped first, since neither changes which file is addressed. A site-root /SUPPORT.md no longer passes. GitHub does not resolve a site-root path against the repository, so that spelling was a broken link the check was accepting; nothing in this repository uses it. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
This was referenced Sep 17, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Prepares the reviewable v1.4.2 (Board clarity) release source change per the work order on #952, based on
mainat6501feb4.pyproject.toml,src/code_mower/__init__.py); committed package manifest regenerated to match exactly.docs/v142-release-notes.mdanddocs/v142-qualification.md;docs/release-history.mdandCHANGELOG.mdupdated, with v1.4.1 marked published and the v1.4.2 entry listing what's actually shipping (Board: manage persistent services and reject stale keepalive bindings during release restart #961'sboard service/board stop --repo, and Board: produce exact local work observations #999/Board: implement the work-first Now, Timeline, Releases, and Health views #1000/Board: manage persistent services and reject stale keepalive bindings during release restart #1001/Board: connect provider-neutral remote lifecycle observations #1002/Board: qualify independent head-bound evidence and session visibility #1003 → Board: integrate head-bound evidence and qualify local and hosted session visibility #951's merged local-evidence code).docs/current-state-and-roadmap.md's Board section refreshed to reflect merged PRs (Board: manage persistent services and reject stale keepalive bindings during release restart #1001, Board: qualify independent head-bound evidence and session visibility #1003) instead of stale "drafts"/"awaiting" language.tests/test_release_v141.pyare untouched (confirmed byte-identical tomain); Graphify facts restored to v1.4.1/Release: qualify and publish optional Graphify context in v1.4.1 #915 where they were mis-attributed to this release.docs/pypi-release.md's post-merge runbook moved forward to v1.4.2, and its Board-restart step (15) was reworked to match the actually-verified live inventory (code-mower board list --json: two services, port 5332codemower-ai/code-mower+ one additional private-repo port — not an assumed third port). It now classifies each port's managed-vs-transient posture fromcode-mower board service status(schema-checked, fails closed on any unclassifiable shape) instead of blindlyboard stop-ing a service Board: manage persistent services and reject stale keepalive bindings during release restart #961 manages, restarts a managed port withboard service restart --replace, and stops a transient one by--repo+--porttogether.shasum).tests/test_release_v142.py(version/runbook/Board-inventory/upgrade-rehearsal/roadmap-fact regressions) and extendedtests/test_release_hygiene.pyto match.Test plan / evidence (fresh venv, unittest's own exit code)
python -m unittest tests.test_release_v142→ 20/20 passed, exit 0.python -m unittest discover -s tests -p "test_release_v14*.py"→ exit 1 overall, with exactly one failure:test_release_v141.InstalledPromptPackTests.test_literal_starter_and_explicit_config_walkthrough. That file is confirmed byte-identical tomain(as aredocs/v141-release-notes.md,docs/v141-qualification.md,docs/v140-release-runbook.md); the failure is this sandbox's own/tmp→/private/tmpsymlink affecting an installed-module-path assertion, not a regression from this change.python -m unittest discover -s tests -p "test_release_hygiene.py"→ 361/361 passed, exit 0.ruff check .,scripts/privacy_scan.py,scripts/guard_package_workflows.py,python -m code_mower.migration release-readiness --json(status: pass) — all clean.c3af1dd230c64996a1ced08427c8b881ce8f6c9a: PASS, zero P0/P1/P2.Remaining orchestrator-owned actions (not performed by this PR)
board service restart/board stop+serve against the reconciled live inventory, andboard doctorverification (step 15).Closes #952.
Co-Authored-By: Claude Sonnet 5 noreply@anthropic.com