Repository navigation
Fix release integrity hash and enforce clean-wheel canonical workflows - #15
Conversation
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 7611775442
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".
| "--out", str(canonical_out)], | ||
| cwd=Path(temp), env=env, check=True, timeout=300, | ||
| ) | ||
| results = json.loads((canonical_out / "canonical-results.json").read_text(encoding="utf-8")) |
There was a problem hiding this comment.
Update the mocked sdist test for the new artifact read
When the core jobs run the full unittest suite, test_check_uses_extracted_cwd_and_isolated_installed_python mocks every subprocess.run call but never creates canonical-results.json, so this new read raises FileNotFoundError before the test's existing three-call assertions are reached. I reproduced this with python -m unittest framework_v2.tests.test_sdist_tutorial_check -v; update the fixture/mock to produce the canonical artifact and account for the additional subprocess invocation so both core matrix jobs can pass.
Useful? React with 👍 / 👎.
| if results.get("readiness") != "RESEARCH-ONLY" or results.get("pit_guarantee") is not False: | ||
| raise ValueError("canonical research boundaries changed") |
There was a problem hiding this comment.
Validate research boundaries on every workflow result
This only checks the document-level readiness and pit_guarantee fields, which examples/canonical_recipes.py writes as unconditional literals; therefore, if any of the four persisted workflow envelopes changes to PAPER-READY or pit_guarantee=true, this release gate still passes as long as its status remains COMPLETED. Check these fields on each results[workflow] entry so the clean-wheel run actually enforces the documented per-result research boundary.
AGENTS.md reference: AGENTS.md:L3-L6
Useful? React with 👍 / 👎.
PR #14 changed package metadata without refreshing its LF-normalized migration hash, causing the Release Foundation runtime integrity test to fail. Refresh the pyproject hash while retaining every integrity assertion.
The package job now also executes all four canonical recipes from the safely extracted sdist with the clean wheel interpreter in isolated mode, verifies the installed module origins and checks each persisted workflow status and research boundaries.
Validation: the previously failing migration-hash test passes; the new clean-wheel gate passes all four workflows and the existing 14 extension tests. Full Linux/Windows and packaging CI must pass before merge.