[codex] Fix suite session recovery - #98
Conversation
…e-london-shoes 修复语言的起源和伦敦鞋子
…q25-answer 修复交易的本能第25题答案
…field-feedback 修复Katherine Mansfield第6题答案
…explanation 修复羊毛产业的历史解析
- 依据「八月高频表格(2).xlsx」的「八月高频文章」本月频次更新 85 条 frequency - A 组(在八月表内不一致)更新为八月本月频次；B 组按「全部文章」表频率；C 组(疑似过度标注)改为 low - examId 保持不变；同步重建 core-foundation.bundle.js
![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: c5c71d265c
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".
k-undurkhaan-2
left a comment
k-undurkhaan-2
left a comment
There was a problem hiding this comment.
Found three additional recovery regressions not covered by the current tests: a stale close watcher can detach a newly resumed window, a permanently invalid snapshot can trap suite mode in an endless retry loop, and the close path reports persistence success even when neither storage backend accepted the snapshot. Details are inline.
|
已经针对审计出的八个comment进行了修复并提交，请PR owner在本地环境执行对抗审计 @dijia702 |
…tmp-migration 1.升级v2数据层 2. 草稿与提交后注释持久化 3. 阅读显示控制
* fix vocabulary list for v2 data layer * fix legacy progress envelope handling
196aa17 to
738de54
Compare
k-undurkhaan-2
left a comment
k-undurkhaan-2
left a comment
There was a problem hiding this comment.
The three previously open recovery fixes now pass revalidation. One new P2 persistence race remains.
…ock, completed close - G1: converge a recovered active-complete multi-suite session before a new run - G5: keep window-WAL fallback when a durable recovery entry exists but is corrupt - F3: set multi-suite recovery write-block on stale/group-conflict and short-circuit - H1: mark a completed suite exam close as completed, not interrupted Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
k-undurkhaan-2
left a comment
k-undurkhaan-2
left a comment
There was a problem hiding this comment.
Re-reviewed 48a7b4b. The previous in-flight draft race remains reproducible. This update also leaves two P2 recovery/ownership failures: a preserved multi-suite WAL cannot pass its next durable CAS after corrupt-durable fallback, and delayed completed-suite teardown can detach a newer ordinary practice session.
|
@codex review |
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 617b82221e
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".
k-undurkhaan-2
left a comment
k-undurkhaan-2
left a comment
There was a problem hiding this comment.
Found two additional recovery failures beyond the existing Codex cross-tab ownership finding. Both are reproducible with focused adversarial cases: exact cleanup misses the fallback-created recovery identity, and unsafe-revision repair is applied only to multi-suite recovery.
k-undurkhaan-2
left a comment
k-undurkhaan-2
left a comment
There was a problem hiding this comment.
Two new P2 ownership races remain on 6a6e2c8: copied sessionStorage is not an exclusive tab lease, and a startup ownership failure is not propagated to the enclosing openExam flow. Both were reproduced with focused adversarial harnesses; the existing regression suites still pass.
k-undurkhaan-2
left a comment
k-undurkhaan-2
left a comment
There was a problem hiding this comment.
Re-reviewed e58f9ba. The two previously reported ownership races pass revalidation. One new P2 startup-recovery lock-order race remains; details are inline.
|
@codex review |
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: e58f9bab96
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".
k-undurkhaan-2
left a comment
k-undurkhaan-2
left a comment
There was a problem hiding this comment.
Reviewed the e58f9ba..bd6fe43 delta for merge blockers. Both addressed recovery defects pass independent revalidation, the affected regression suites and generated-bundle parity checks pass, and I found no new blocking regression in this update. Per review scope, systemic legacy debt is left for a separate PR; no new inline comments.
状态说明：当前 PR 的定位与 base 分支不匹配核心修复有效
当前 PR 的定位与 base 分支不匹配
数据层同步长出了配套原语（ 租约、fence、tombstone、exclusiveGroup、CAS——这些是**多写者（multiple writers）**场景的并发控制设施。而
建议：本 PR 改指向
|
5 participants
Summary
sessionStorageorlocalStorageand reopen the current passage after a main-page refresh.suiteSessionIdacross navigation andfile://fallback paths.Root cause
The parent page only mirrored suite state after later events, while child-window closure called
_abortSuiteSession, which cleared the persisted snapshot. Refreshes andfile://storage differences could therefore lose the active suite.Validation
file://suite E2E flow passes.The repository's existing static CI baseline still reports unrelated failures in CSS convergence, settings-button coverage, on-demand test mocks, service-facade CSS checks, and IndexedDB shadow cleanup.