Skip to content

test(chat): restore evidence subject before linked recovery - #7620

Merged
loopx-agent merged 1 commit into
loopx-project:mainfrom
xyx2002OvO:codex/ci-team-evidence
Oct 29, 2026
Merged

loopx-agent merged 1 commit into
loopx-project:mainfrom
xyx2002OvO:codex/ci-team-evidence

Conversation

@xyx2002OvO

@xyx2002OvO xyx2002OvO commented Oct 29, 2026 •

Copy link
Copy Markdown
Contributor

Summary

  • restore the original Todo-scoped validation subject before the linked recovery journey
  • keep the deliberately mismatched output version while avoiding contamination from the preceding request-validation case
  • change only the durable browser smoke; product behavior is unchanged

Failure and ownership

team-evidence-return-smoke.mjs first replaces validation with a request-level fixture whose todo_id is null. The later linked-recovery case reused that mutated object, so the component correctly rejected the evidence revision and the expected verifier-gap text never appeared. The same timeout at line 133 failed both #7603's chat-bundle-browser (2) job and its packaged Release Artifacts / build job.

This PR extracts only the second, team-evidence-specific commit from #7604 onto current main@652f503842609e948e8ff832ae3977f23f79f28d. It intentionally excludes #7604's delegation-search and team-feedback fixture changes; #7599 already owns the delegation-search fix, while the remaining team-feedback isolation stays with #7604's owner.

Validation

  • hosted chat-bundle-browser (2): passed, including Qualify packaged team evidence and feedback recovery
  • hosted Release Artifacts / build: passed, covering the same packaged team-evidence path that failed on fix(runtime): enforce private IO and preserve Windows long paths #7603
  • hosted dashboard-browser (2): passed with the existing M2 budget unchanged
  • hosted DCO, dependency review, change classification, kernel static, chat bundle and macOS desktop checks: passed
  • npm run build:chat: passed locally
  • packaged team-evidence-return-smoke.mjs: passed locally through checked outputs, linked recovery, evidence loss/revocation/restoration, pagination, mobile and no-execution assertions
  • node --check apps/presentation/dashboard/smoke/team-evidence-return-smoke.mjs: passed
  • diff-scoped standard premerge canary with an empty isolated Windows user profile: 5 selected checks passed, 0 failures
  • git diff --check: passed

The repository's POSIX-style combined npm script cannot be invoked directly by Windows cmd; its unchanged team-feedback half also observes this machine's pre-existing dual-default-runtime conflict. Neither condition is caused or masked by this one-file change. Other shared matrix jobs remain pending.

No merge is requested by this submission.

Signed-off-by: Tartar <xiaoyx67@mail2.sysu.edu.cn>

@loopx-agent loopx-agent left a comment

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Reviewer: model_agent · gpt-6.1-sol · OpenAI · runtime_reported · max

动机 / Motivation

维护团队证据阅读与发布验证的开发者。 先验证普通原请求,再回到同一Todo的关联证据恢复时,旧fixture残留null任务身份而被页面正确拒绝;恢复原Todo验收基线后,原来的关联恢复与缺失独立验收提示可再次被完整验证。 本头恢复了这条稳定浏览器回归的原始验证对象,并保留错版本拒绝、证据失联清除、恢复、分页、390px和禁止新增执行的全部断言。 该PR仅修复测试数据污染,不修改产品组件、HTTP契约、身份判断、权限、默认行为、等待上限或真实模型验收。

改动思路 / Approach

修复放在已有smoke的局部fixture切换,保留生产owner及独立负例,不增加机制。仅恢复既有Todo验证主体;生产行为、权限、默认值和等待预算不变。前面的普通原请求案例把验收来源改为 request_validation,其 todo_id 为 null;而 GoalTeamEpisode.trace 必须把后续读回与当前 Todo 的 operation/request/Agent/产物准确绑定。直接回到同文件已有 currentValidation,比延长等待、删断言、给产品加fallback或再造fixture更合适;此次仍保留故意错版本,未改变任何预期结果。

具体改动 / Changes

关键代码讲解

currentValidation(line57)保留原 Todo 验收来源、时间和两份产物版本,是后续独立案例的既有基线。line103 的 validation 重建改为展开这份基线;前一请求验收案例自身未改,report.json 的 f-digest 仍用于证明另一版本不能成为当前检查记录。随后原 recheck、核心失联/恢复、关联读回、独立验收缺口和对照选择断言全部实际运行。

规范为 docs/reference/local-delegation.md,固定 revision 652f503842609e948e8ff832ae3977f23f79f28d:exact-version 对应准确版本及缺失检查记录的正反例;independent-gap 对应“任务接受/采用不能代替独立验收”;reader-recovery 对应核心撤回、下游失联仍保留当前纠偏和显式恢复。三项均 implemented。净diff仅此smoke的2增1删,产品组件零变化;复用现有来源标识,无新vocabulary。

对主干的风险 / Risks

主要风险是为了通过测试而掩盖主体或产物不匹配。本头保持原产品精确匹配,错版本仍要求时间/版本记录缺失,失联仍撤回对应证据,恢复仍核对原主体,且没有放宽超时或任何断言。PASS:exact head 完整 packaged team-evidence-return 与相邻 team-feedback-recovery 均通过;node syntax、RFC bilingual、diff hygiene及4项直接/5项选中premerge检查通过,零失败、跳过与人工hold。所有检查使用隔离HOME;未查询CI。 相邻反馈恢复也在同样隔离环境完整通过;这与仅挑选一个happy path不同。原基线的line133失败保留为fixture污染的实际观察,不改写成通过;#7599的委派搜索与#7604的其他工作未复制进来。

我的整体评价 / Verdict

APPROVE,范围完整且可独立回退。长期Agent效果不适用于这项测试修复;用户实际页面及操作保持原样。净diff仅改变smoke中验证对象的展开源和注释,没有渲染、布局、文案或首屏变化。真实模型纠偏和已安装Native App仍不由合成浏览器fixture证明。这个小修复恢复既有回归的价值,不产生新的用户配置或产品兼容成本;整版发布的后续门仍独立完成。

English verdict: APPROVE f869d7b500f82a26e5a6b1341abd140c681d8a07 — restore the original Todo validation subject without weakening version, recovery or no-execution assertions; packaged positive/negative journeys and local premerge checks passed. CI was not consulted.

@loopx-agent
loopx-agent merged commit 6fa2cf5 into loopx-project:main Oct 29, 2026
36 of 42 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants