Repository navigation
ci: report six unit results from one workflow and retire the legacy PR gates - #2450
Merged
Merged
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: 9c54f1c263
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".
joeharris76
enabled auto-merge
September 29, 2026 17:08
…R gates Add ci.yml with always-reporting core, explorer, results-data, docs, landing, and tooling results driven by a path ownership map, and retire the workflows those results replace. The docs workflow keeps develop baseline capture and the release Pages deploy. Ruleset drift pins move to the new contexts and a smaller merge-queue build concurrency.
…udget Add a merge_group-only dist-artifact job that builds the wheel and sdist, hashes them into SHA256SUMS, and uploads dist-<sha> for 90 days so the release workflow can publish the exact queue-tested files. Include it in the core unit result expectations. Trim the branch policy paragraph in AGENTS.md to fit the 16000-byte instruction budget.
The flip deletes results-explorer-browser.yml; the Chromium full suite now runs as explorer-e2e inside ci.yml. Fixes the content-guard relative-link failure on the flip PR.
- Give the submission validator sync step its required branch-copy arguments (fetch published-results, compare the two copies). - Replace the stale _project fixture path in the classifier test with a placeholder stem the project-reference gate ignores. - Format the classifier test files with ruff.
Point the standardized command guards and the lint parity contract at the ci.yml jobs that replace pr.yml. Resolve repo root from the test file location instead of the process cwd so the guards survive a cwd leak. Cover the new ci.yml guard steps with local equivalents or documented hosted-only exemptions.
The flip deletes the standalone base-guard workflow, so a stacked PR would otherwise report green unit results without the loud wrong-base failure. Move the supported-base check into ci.yml as a base-guard job wired into the tooling result, run the public-site visual regression lane on all PR bases, restore the auto-merge revocation workflow that the deletion over-reached into, and verify bundled dbgen binaries emit clean framing in the correctness gate.
…e CI workflow Codex review found two gaps the flip left behind. The deleted tpch-binary-framing matrix was the only pre-merge macOS/Windows execution of the bundled dbgen framing guard; folding one Linux step into plan-capture-gate did not preserve that coverage. Restore it as a standalone macos/windows matrix job wired into the core unit result, matching the exemption the parity test already anticipates. Restore the deleted stacked-PR guard test retargeted at the new base-guard job in ci.yml, including a pin that a failing guard fails the tooling unit, and exempt the branch-identity step from local parity with a reason.
The results-data unit compared the submission validator against the live published-results copy on every results-data change, so any standing drift between the branches failed unrelated changes. Run the sync check only when the change touches the validator or its checker; submission-validator-drift-check.yml still reports standing drift weekly and on develop pushes. Also point the complexity-policy pin at ci.yml, which replaced pr.yml.
The single CI workflow retires the publication lane workflows and the lane-isolation-check Make target they fed. Remove the test that invoked the target and the parity-doc paragraph that described it as a local equivalent.
The ledger listed workflows and tests the flip deletes and lacked the new unit-result tests, so the ledger coverage tests failed. Remove the deleted rows and add the new ones. gitignore-lint has no equivalent in ci.yml, so keep it as a standalone workflow. Drop the skill-integrity step's reference to the deleted path classifier test.
joeharris76
force-pushed
the
feat/dev-loop-v2-flip
branch
from
September 29, 2026 18:38
2b11a26 to
6d639e5
Compare
The develop ruleset now requires core, explorer, results-data, docs, landing and tooling. The landing readiness check, the green-but-unmerged sweep and the soundness drain digest still asked for the retired contexts, so they would have reported every PR as not green or refused to arm it. Update the required lists and their fixtures. The delta guard in ci.yml restores a per-commit fast-lane count that only the deleted post-merge workflow wrote, so it would fail closed on every PR. Add a small develop-push workflow that records the count. The nightly sweep no longer checks the retired post-merge workflow.
The legacy PR, browser, ruleset-drift, post-merge and refresh-shadow workflows are gone, but docs, comments and a few scripts still named them or the old required checks. Point them at ci.yml and the six unit results, and rewrite the sections that described behavior that no longer exists (the post-merge sweep and gap detector, the auto-revert labels, the browser gate wiring, the base guard). Two scripts read the deleted file and would have failed: the slow-test coverage report and the medium-test timeout lookup in the PR metrics. The metrics script now resolves the required contexts per head, so PRs merged before the change still measure against the checks that gated them.
pip-audit flags pyjwt 2.13.0 for CVE-2026-102274 with a fix in 2.14.0. pyjwt is a transitive dependency, so refresh the lock to 2.15.1.
joeharris76
disabled auto-merge
September 29, 2026 21:58
joeharris76
added this pull request to the merge queue
Sep 29, 2026
github-merge-queue
Bot
removed this pull request from the merge queue due to failed status checks
Sep 29, 2026
joeharris76
added this pull request to the merge queue
Sep 29, 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
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
Replace the legacy PR-gate and post-merge workflows with one
ci.ymlthat reports six always-on unit results (core,explorer,results-data,docs,landing,tooling). Untouched units pass immediately; touched units must succeed in every required job. Includes the merge-groupdist-<sha>artifact upload (wheel, sdist, SHA256SUMS) for the release workflow, the soundness-flag review gate, post-flip drift pins, and updated agent instructions.Soundness review:
External reviewer: codex
Review output: #2444 (comment)
All Critical/High findings resolved.
Validation