Skip to content

test(sweep): make completion-order regression deterministic - #2033

Draft
seonghobae wants to merge 3 commits into
mainfrom
fix/agent-sweep-completion-order-regression
Draft

test(sweep): make completion-order regression deterministic#2033
seonghobae wants to merge 3 commits into
mainfrom
fix/agent-sweep-completion-order-regression

Conversation

@seonghobae

Copy link
Copy Markdown
Contributor

Status

Proposed / Draft. This bounded successor preserves the remaining agent-mention sweep test delta from historical #1231 without restacking its conflicted multi-owner branch.

Root cause

list_recent_pull_requests already submits bounded repository fetches concurrently and yields with concurrent.futures.as_completed. The current-main regression instead claimed results remain repository ordered and asserted first, second without controlling completion. It therefore tested thread scheduling and contradicted the latency boundary: a completed repository must not wait behind a slow sibling.

RED → GREEN

  • RED 570da463b0ce7f4837727a91557187624a7b37db: introduce a deterministic two-repository fixture that blocks first while retaining the obsolete first, second expectation. Production necessarily yields second first.
  • GREEN de96b8b46143fe63d8fec1929b5739a4babee8c4: assert second, first, rename the test to the completion-order contract, and update CHANGELOG, doctoring, and docs/product-technical-gap-baseline.md.

Boundary

  • Production scheduler, retry, credential, merge, and scanner code is unchanged.
  • Worker ceiling remains exactly two in the fixture.
  • No dependency, abstraction, timeout policy, force push, or bypass.
  • Exact-head hosted Checks and independent review remain required before integration.

Verification authority

Branch was created from protected main@78a4937c684a54ca8e415822c913742f41c6efc4; both commits are normal descendants. Predecessor-head evidence is not inherited.

@coderabbitai

coderabbitai Bot commented Sep 8, 2026

Copy link
Copy Markdown

Important

Draft PR not reviewed

Draft PRs are not automatically reviewed by default.

  • Trigger a manual review

To automatically review draft PRs, update your CodeRabbit configuration:

reviews:
  auto_review:
    drafts: true

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

Comment @coderabbitai help to get the list of available commands.

Copy link
Copy Markdown
Contributor Author

Exact-head receipt:

  • exact head: de96b8b46143fe63d8fec1929b5739a4babee8c4
  • protected base: main@78a4937c684a54ca8e415822c913742f41c6efc4
  • ancestry: 2 ahead / 0 behind
  • delta: 4 files; production source unchanged
  • lifecycle: Draft / Proposed; mechanically mergeable
  • review threads: 0
  • fresh runs: Runtime Quality 34181386071, Agent Mention Router Quality 34181386104, Python Security 34181386113, Security Scan 34181386092, SAST 34181386099, CodeQL 34181386094; all remain queued

No predecessor evidence, bypass, auto-merge, or synthetic status is used.

Copy link
Copy Markdown
Contributor Author

Exact-head verification receipt for de96b8b46143fe63d8fec1929b5739a4babee8c4: Agent Review Runtime Quality run 34181386071, Agent Mention Router run 34181386104, Security Scan 34181386092, and SAST Semgrep 34181386099 completed successfully. Python Security and CodeQL PR remain queued, and no independent review is present. The PR therefore remains Draft/Proposed; these successes are exact-head evidence but not merge authorization.

Copy link
Copy Markdown
Contributor Author

CodeQL exact-head follow-up: run 34181386094 is in the designed two-phase state. Shards 101922706890 and 101922706920 failed only with DISPATCH_OUTCOME=success, VERDICT_STATE=pending; the coordinator job 101924039777 is queued pre-runner with steps=null. This is not CodeQL analysis failure or GREEN. Do not manually rerun the shards before the coordinator publishes authenticated terminal evidence.

seonghobae added a commit that referenced this pull request Sep 8, 2026
Preserve protected-main base/run receipt binding while moving multilingual wake authority to one post-matrix failed-jobs request. Record live #2033 HTTP 403 reproduction and focused contract evidence.

Copy link
Copy Markdown
Contributor Author

Exact RCA handoff: dispatch run 34182987578 completed both CodeQL scans and published codeql-dispatch/python=success and codeql-dispatch/actions=success. Python woke required job 101922706890 first; actions job 101927877852 then failed its wake because required run 34181386094 was already running (HTTP 403). Canonical owner repair .github#2032 has been non-force restacked at 2268986b3b5f1ee6cd5ad738284c480f653b9078: one post-matrix coordinator validates the full failed set and calls rerun-failed-jobs once, while preserving #2028's exact base/run receipt binding. #2033 remains Draft; clean dispatch statuses do not replace terminal exact-head required checks.

@seonghobae seonghobae added bug Something isn't working priority: medium Normal-priority or P2 work type: bug Defect or incorrect behavior labels Sep 8, 2026 — with ChatGPT Codex Connector

Copy link
Copy Markdown
Contributor Author

Restack onto protected main@7fd571dbcdbae6acf29d8f4ee704d7ba6297e4db with non-force git merge --no-ff origin/main.

  • previous head de96b8b46143fe63d8fec1929b5739a4babee8c4 remains an ancestor
  • new head d343f564f813345091e27ec841940b93abb9fc58
  • conflicts: none
  • focused: PYTHONPATH=. python3 -m pytest tests/test_agent_mention_sweep_regressions.py → 9 passed

Left Draft. Fresh exact-head checks required.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

bug Something isn't working priority: medium Normal-priority or P2 work type: bug Defect or incorrect behavior

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant