[PR] test(desktop): retire unreliable held-thumb geometry probe [maka]

["M4n5ter (via GitHub)" <gi...@apache.org>]

M4n5ter opened a new pull request, #5346:
URL: https://github.com/apache/maka/pull/5346

   <details open>
   <summary>English</summary>
   
   ## Summary
   
   Remove `scroll-geometry.spec.ts` and its E2E budget entry. This is the CI cleanup below #5345; it changes test coverage, not production scrolling.
   
   The probe treats a fixed 50 ms after synthetic thumb movement as a stationary baseline while Host history can arrive concurrently. Diagnostics observed 1.3125 px and 4 px failures without an application `scrollTop` write; browser scrolling continued after the sampled baseline. A no-publication control still showed a native tail step. Pixel-rounding the injected input did not give a reliable fix. These observations do not establish a production anchoring defect, and making the probe deterministic would require additional publication/input synchronization.
   
   Keep the existing Host paging E2E, origin-prepend and cold-content browser geometry checks, and scroll-authority tests. Remove stale references to the retired probe.
   
   ## Coverage tradeoff
   
   The remaining tests are **not equivalent**: there is no longer a real Electron held-thumb/across-origin/history-publication scenario enforcing a 1 px bound on every frame. In particular, the retained wheel paging E2E has a 40 px range-boundary allowance. This PR accepts that gap instead of retaining a gate that cannot reliably separate input completion from content-induced movement. It does not claim to fix native scrollbar drift.
   
   ## Verification
   
   - Final stack: complete local Electron E2E — **38 passed** (6.4 minutes, 1.0 GB peak). Reordering the stack preserved the tested tree exactly.
   - 38 scroll-authority and chat-scroll tests passed.
   - E2E budget: 38 tests in 21 files; checker tests passed.
   - Biome and diff checks passed. Production scrolling code is unchanged.
   - WorkHub UI evidence is in #5345, including screenshots in both language sections.
   
   ## AI use
   
   - [ ] No generative tool made a substantive contribution
   - [x] Generative tooling made a substantive contribution
   
   Tool(s) and scope: OpenAI Codex investigated the failure, audited the retained coverage, removed the probe, and ran local validation.
   
   ## Checklist
   
   - [ ] Tests cover the change and fail without it — intentional deletion; the remaining coverage and lost scenario are documented above.
   - [x] Lint, format, typecheck and the affected suites pass locally
   
   Does this PR entail a change in behavior?
   
   - [x] Yes — CI no longer runs this probe; production behavior is unchanged.
   - [ ] No
   
   </details>
   
   <details>
   <summary>中文</summary>
   
   ## 改了什么
   
   这是 #5345 底层的测试清理 PR：删除 `scroll-geometry.spec.ts` 及其 E2E 预算条目，清理文档和注释中的旧引用，不改生产滚动逻辑。
   
   这条测试在模拟拖动后固定等 50ms，就把当前位置视为已经静止；但 Host 历史可以同时到达。诊断中出现过 1.3125px 和 4px 的失败，期间应用没有写入 `scrollTop`，浏览器还在调整滚动位置。暂不发布历史的对照也记录到了原生输入尾帧；对输入坐标取整同样没得到可靠修复。这些结果不足以证明产品锚定有缺陷。若继续保留，需要额外控制历史发布与原生输入的时序，成本超过当前检查能提供的信心。
   
   真实 Host 分页、原点补页、冷内容几何和滚动状态测试继续保留。
   
   ## 明确的取舍
   
   删除后，不再有“真实 Electron 滚动条持按跨原点、同时补入历史”的逐帧 1px 联合检查。其他测试不能完全替代它；其中保留的滚轮分页 E2E 允许范围切换时有 40px 位移。本 PR 接受这个覆盖缺口，不把删除测试说成修复了原生滚动问题。
   
   ## 验证
   
   最终 stack 的完整本地 Electron E2E：**38 项全部通过**，耗时 6.4 分钟、内存峰值 1.0 GB；调整分支顺序后，代码树与通过测试的版本完全一致。38 项滚动 authority 与 chat-scroll 测试、E2E 预算检查及 Biome 均通过，预算为 21 个文件、38 项 E2E。WorkHub 界面截图放在 #5345，中英文两区都有。
   
   ## AI 使用
   
   OpenAI Codex 参与诊断、保留覆盖审计、测试删除及本地验证。检查项以英文区为准；此 PR 保持 Draft。
   
   </details>
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: commits-unsubscribe@maka.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscribe@maka.apache.org
For additional commands, e-mail: commits-help@maka.apache.org

Re: [PR] test(desktop): retire unreliable held-thumb geometry probe [maka]

["M4n5ter (via GitHub)" <gi...@apache.org>]

M4n5ter merged PR #5346:
URL: https://github.com/apache/maka/pull/5346


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: commits-unsubscribe@maka.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscribe@maka.apache.org
For additional commands, e-mail: commits-help@maka.apache.org