Repository navigation
test(codex): preserve unsupported generation against weak native evidence - #310
Conversation
Native Codex 0.162.1 exposes no per-incarnation identity and no atomic original-generation precondition for queued input: the writer lock file is empty, sessionId survives resume, thread/queue/add takes only the thread, input and a client message ID, and queued rows are consumed by whichever process has the thread loaded. Keep Codex generation unsupported and add negative fixtures proving that stable writer evidence, lock content, queue and client message IDs, direct-input capability, sessionId, turn IDs and writer restarts never mint a target generation or start any effect. Document the pinned source evidence.
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. |
|
Exact head aaf75af additionally isolates the newer CoreFoundation preference reader from host MDM state in generation fixtures. Existing unsupported/no-effect assertions are unchanged. Five focused fixtures pass against this branch and the PR311 runtime; omitting the isolation reproduces three host-layer failures in the controlled mutation check. This is test isolation, not native generation support. Pending current-head CI and review; #180 remains open. |
|
@codex review |
|
Codex Review: Didn't find any major issues. Already looking forward to the next diff. Reviewed commit: ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
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". |
|
@codex security review |
🛡️ Codex Security ReviewSecurity review completed. No security issues were found in this pull request. Reviewed commit: Only the user who started this review can view the report in Codex. ℹ️ About Codex security reviews in GitHubThis is an experimental Codex feature. Security reviews are triggered when:
Once complete, Codex will leave suggestions, or a comment if no findings are found. |
…ion-unsupported # Conflicts: # docs/i18n-manifest.json
|
Updated to current main57c42d4 after #305/#311 merged. Exact integration head9780bdb89f019a9302088b7eeb9591cfce8a8c18. Only merge conflict was docs/i18n-manifest.json CLI-reference digest; all4locale source paragraphs preserved, actualEnglish SHAa1aae5b881f4aa939c995d7bc06819d04f808da3b31fd0d5da66c817e1b47e82 recomputed. Generation5/docs61/generator/diff pass locally; no product behavior or unsupported-generation expectations weakened. Waiting current-head CI/reviews before normal squash merge; #180 remains OPEN. |
|
@codex review |
|
@codex security review |
🛡️ Codex Security ReviewSecurity review completed. No security issues were found in this pull request. Reviewed commit: Only the user who started this review can view the report in Codex. ℹ️ About Codex security reviews in GitHubThis is an experimental Codex feature. Security reviews are triggered when:
Once complete, Codex will leave suggestions, or a comment if no findings are found. |
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 9780bdb89f
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".
|
@codex security review |
|
Codex Review: Didn't find any major issues. Chef's kiss. Reviewed commit: ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
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". |
🛡️ Codex Security ReviewSecurity review completed. No security issues were found in this pull request. Reviewed commit: Only the user who started this review can view the report in Codex. ℹ️ About Codex security reviews in GitHubThis is an experimental Codex feature. Security reviews are triggered when:
Once complete, Codex will leave suggestions, or a comment if no findings are found. |
1 participant
Summary
Refs #180; does not close it. Preserve the shipped Codex
targetGeneration=unsupportedboundary and add five offline negative fixtures. No product runtime, schema, dependency, CLI flag or package version changes.sessionId, revisions and turn IDs do not mint a generation.Type
Checklist
cc_peer.pycommands are not used as evidence.Verified against
Pinned native source: openai/codex rust-v0.162.1,
092d3acd6bec3e3a14bdc7e7a2810ab628ab759d. Static source inspection, not actual native restart or delivery acceptance. Parent implementation report ran Python 3.13 extras: 1109 tests OK (19 skipped), without new live model calls. The Python 3.9 default suite's missingpackagingfailure also occurred on the base; this PR does not repair that unrelated environment.Direct source references: queue parameters, queue processor, lock acquisition, resume identity, active-turn precondition.
Independent review
Claude Code 2.1.300 with requested and response-confirmed
claude-opus-5-5; Codex CLI 0.162.1 with isolated default model (actual model name not exposed in captured metadata), read-only sandbox and ephemeral execution. Both non-interactive, separate processes, same immutable stdin, no reviewer tools/tests/network/edits, no cross-sharing of reviewer conclusions.Initial joint input SHA256:
2d04e0a30998dcf924a447298e7b80806f5682f2e4c793a3325d5cbfe90ef306. Follow-up #180 input SHA256:e85be5543fcd15a73b36198cf37b2f58f11aa4bfd08bf956833489404e1d80ef. All four review processes completed with exit 0 and valid final responses; immutable inputs retained their hashes. The follow-up reviewed environment isolation and mock cleanup with additional pinned source excerpts. Required scope/lock wording corrections were subsequently applied in all four locales. Recommended HOME/USERPROFILE/CODEX_HOME fixture pinning, mock invocation checks and synthetic-test naming were also applied and checked by the coordinator. These final review-driven edits passed local tests but were not independently rerun; current-head GitHub review remains a separate gate. Static review is not native/platform/receipt proof.Notes
expectedTurnIdis promoted into this guarantee.Integration follow-up (2026-10-16)
Updated onto protected main 57c42d4; resolved only the combined CLI-reference manifest hash. Automatic P2 review was valid: the earlier success test replaced the producer under test. Follow-up 88e19f9 instead exercises the actual CLI/adapter/queue producer, mocking the native subprocess and fixture prerequisites. Python3.13 and3.9 focused five tests pass; a real-producer-result in-memory mutation is detected. This is a bounded local non-wake fixture, not all-path/native/platform acceptance or proof of zero ambient metadata reads.
Independent follow-up: same frozen input SHA256
4c736e9f0eb346b6b58c4de0cbed48ace6de644b6c50101dc1853fa56830fa7c; Claude Code2.1.301 actualclaude-opus-5-5and Codex CLI0.162.1 isolated default (actual model unexposed), parallel non-interactive/static/no tools, both exit0 with complete results and no mandatory defect. Current-head GitHub checks and reviews remain the merge gate.