Repository navigation
Skip closed and merged PRs before posting wake comments - #141
Conversation
|
Integration update: published head Root reviewed the combined changes and ran 23 focused tests successfully, including all seven mocked wake-state tests plus affected historical/current metadata and existing wake tests. The native published tree exactly matches the verified local tree. PR141's authored scope remains the fresh PR-state guard and its tests; no live poster was executed. New full CI is pending, and no mathematical status changes. |
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. |
|
Batch 389 wake @2f7a5a9 — project-intent eng resume (not research flip) Coordinator: https://cursor.com/agents/bc-01a0cf1e-ebff-78a8-8a7a-9140fd59309a Intent: tip chatgpt/drive-github-hardening-20260919 @ 2f7a5a9; Path C IDLE@0019; lemma_closed=false; scientific effect NONE; no claim promotion Eng resume task: Resume Path C eng advance on this trial PR — additive-only; no lemma_closed flip; no claim/research promotion. Path C intent advance only. Tip-align if base drifted; additive-only; never promote claims/premises/prizes/lemmas; no lemma_closed flip. Trial wake path: https://github.com/d6g8k5htny-coder/trial / workflow |
|
PR140 is merged at The diff against integrated main is now exactly the wake guard and its seven mocked tests. Root ran 33 focused tests on the actual combined tree, including the new owner Batch388–391 checks; all passed. Full five-job hosted CI had already passed on prior guard head |
The wake poster checks living-tip deduplication but does not verify that its target PR is still open. Its static main targets include main PR36, which merged as
534f4fce08be919cb969ca87963c5210148497a1on 2026-09-26. A new upstream tip can therefore cause another wake on completed work. Trial targets can also close after the open-PR listing.This PR adds a fresh
GET /repos/{repo}/pulls/{number}inside_post_or_skip, before comment retrieval or writing. Closed/merged PRs are recorded as skipped. Unknown or malformed PR state refuses the wake, and API errors propagate before a POST is possible. Open-target task bodies and living-tip deduplication retain their existing behavior.The seven new stdlib/mock tests cover closed and merged targets, an inconsistent open/merged response, open-target posting and deduplication, malformed responses, API failures, and a trial PR closing after listing. They block token and network access. The original script failed these tests; the repair passes all seven in normal and optimized Python 3.12.14. Ten selected pytest tests also pass, including three existing wake tests.
git diff --checkpasses.Latest trial main
fb08e5ac8258edbc605539d7a464099f2de832f6was merged ordinarily into the isolated repair, preserving concurrent batch edits. Native GitHub tree/commit publication was fetched back: remote treef0367091c1a599732961fc650ee7b78a190ed6d7exactly matches the locally verified tree. The diff against that main contains only the poster and the new test module.The poster was never executed live. No local token access or API calls occurred during tests. The full trial suite was not run locally because it contains live probes; hosted checks remain pending. This PR is separate from PR140's workflow and historical Intent-test repair. No schedules, token handling, permissions, research status, or scientific claims change.