Skip to content

feat(web-shell): drop the mobile composer's history arrow buttons - #14220

Merged
wenshao merged 6 commits into
mainfrom
fix/web-shell-mobile-composer-history-arrows
Oct 16, 2026
Merged

wenshao merged 6 commits into
mainfrom
fix/web-shell-mobile-composer-history-arrows

Conversation

@pomelo-nwu

Copy link
Copy Markdown
Collaborator

What this PR does

Removes the two one-entry-at-a-time history buttons (↑ / ↓) from the Web Shell touch composer's editing row, together with the composer-core callbacks and the English/Chinese strings that only those buttons used. The rest of the row — the labelled history button shown when a host has not opted into the Add entry, the Shell toggle, keyboard dismiss, and expand — is unchanged, as are the history search panel and every desktop code path.

Why it's needed

On a phone the editing row's two most prominent leading slots were spent on a low-frequency action, adding a full row of visual noise to the screen where focus matters most, while a strictly better history entry point already existed: the searchable history panel, reachable from the Add drawer and from the labelled history button. The touch restore path behind the arrows was also lossy by design — inline @file tag chips are not rebuilt when history is restored into the plain mobile textarea — so the best seat in the mobile UI went to a degraded experience. The issue carries the full rationale and a side-by-side comparison with the Codex mobile composer.

One deviation from the issue text, called out for reviewers: the issue assumed the two navigation callbacks would stay for desktop keyboard navigation. They would not — desktop arrow keys are implemented inline by the editor's ArrowUp/ArrowDown keymap handlers and never called those callbacks, so once the buttons go the callbacks have zero callers. This PR therefore deletes them, including their touch-textarea branches, instead of leaving dead exports; that is what the issue's own "remove the branches if nothing calls them afterwards" instruction reduces to.

Reviewer Test Plan

How to verify

Open the Web Shell on a touch device (or desktop DevTools with touch emulation) and focus the composer: the editing row above the textarea must no longer contain "Previous input" / "Next input" buttons in either toolbar configuration, with and without the host addMenu opt-in. With the Add entry present, tap + then "Input history": the search panel opens, lists past inputs, filters as you type, and tapping a match restores that text into the composer and returns focus without submitting. Without the Add opt-in, the editing row still shows the labelled history button and the Shell toggle, and both stay disabled while the composer is disabled. On desktop nothing changes: with a single-line draft, ArrowUp/ArrowDown still walk prompt history and restore the draft saved before browsing.

Automated: cd packages/web-shell && npx vitest run --config vitest.config.ts (393 files / 10396 tests passed), npx tsc -p tsconfig.json --noEmit (clean), and npx playwright test --config playwright.config.ts --project=mobile-chromium client/e2e/web-shell.composer.mobile.spec.ts (21 passed). The rewritten Playwright test now asserts that every button in the editing row keeps a 44px touch target and stays inside the viewport at 390px and 240px widths, with no horizontal overflow.

Evidence (Before & After)

Before: the issue's first screenshot shows the ↑ / ↓ pair occupying the leading slots of the editing row. After: captured locally with the repository's visuals harness (playwright test --config playwright.visuals.config.ts -g "mobile composer controls", Pixel 7 at 390x844) and reviewed by eye — the editing row renders only the keyboard-dismiss and expand controls, right-aligned by the row spacer, with the leading slots empty while the Add entry is configured; no leftover gap, misalignment, or horizontal overflow at 390px or 240px. The captures land in the git-ignored client/e2e/visuals/output/screenshots/ directory and are not attached here.

Tested on

OS Status
🍏 macOS ✅
🪟 Windows ⚠️
🐧 Linux ⚠️

Environment

The packages/web-shell Vite dev server driven by Playwright, with chromium installed locally via playwright install chromium; the mobile-chromium project provides Pixel 7 touch emulation.

Risk & Scope

  • Main risk or tradeoff: mobile users lose one-entry-at-a-time history stepping, leaving the search panel as the only touch recall path. Mitigated by the panel being strictly more capable (full list plus filter) and already touch-ready.
  • Not validated / out of scope: real-device iOS/Android keyboard behavior (browser emulation only, per the design doc's open questions); the mobile-webkit Playwright project was not run locally because the WebKit browser is not installed here — CI covers it. Desktop editing, daemon routes, and host toolbar opt-ins are untouched.
  • Breaking changes / migration notes: none for hosts. No public API or customization surface changed; the removed callbacks were internal to the web-shell client.

Design documentation updated in both languages: English | 简体中文.

Linked Issues

Closes #14137

中文说明

本 PR 做了什么

从 Web Shell 触屏输入框的编辑操作行中移除逐条翻阅输入历史的 ↑ / ↓ 两个按钮,并一并删除只有这两个按钮使用的 composer core 回调与中英文文案。该行其余部分保持不变:宿主未启用 Add 入口时显示的带标签历史按钮、Shell 开关、收起键盘和展开;历史搜索面板与所有桌面端代码路径也保持不变。

为什么需要

在手机上,编辑操作行最显眼的两个行首位置被一个低频操作占用,给最需要聚焦的界面增加了整整一行的视觉噪音,而一个明显更好的历史入口其实已经存在:可从 Add 抽屉和带标签历史按钮进入的历史搜索面板。箭头背后的触屏恢复路径本身也是有损的——历史恢复到移动端纯文本 textarea 时不会重建内联 @file 标签 chip——因此移动端 UI 最好的位置一直给了一个残血体验。完整理由及与 Codex 移动端输入区的对比见 issue。

有一处与 issue 原文不同的地方,提请评审注意:issue 假设这两个导航回调会保留给桌面端键盘导航。实际上不会——桌面方向键由编辑器的 ArrowUp/ArrowDown keymap 处理函数内联实现,从不调用这两个回调,所以按钮移除后它们将没有任何调用者。本 PR 因此连同其 touch-textarea 分支一起删除这两个回调,而不是留下死代码;这正是 issue 中"若之后没有移动端调用者就移除这些分支"这一指示的最终归约。

评审测试计划

如何验证

在触屏设备(或桌面 DevTools 的触摸模拟)上打开 Web Shell 并聚焦输入框:输入框上方的编辑操作行在两种工具栏配置下(宿主启用与未启用 addMenu)都不应再出现 "Previous input" / "Next input" 按钮。存在 Add 入口时,点击 + 再点 "Input history":搜索面板打开、列出历史输入、随输入过滤,点击某条匹配会把该文本恢复进输入框并交还焦点,且不会发送。未启用 Add 时,编辑操作行仍显示带标签历史按钮和 Shell 开关,且输入框禁用时二者同样禁用。桌面端无变化:单行草稿下 ArrowUp/ArrowDown 仍能翻阅历史并恢复开始翻阅前的草稿。

自动化:cd packages/web-shell && npx vitest run --config vitest.config.ts(393 个文件 / 10396 个用例通过)、npx tsc -p tsconfig.json --noEmit(无错误)、npx playwright test --config playwright.config.ts --project=mobile-chromium client/e2e/web-shell.composer.mobile.spec.ts(21 个通过)。改写后的 Playwright 用例现在断言编辑操作行内每个按钮在 390px 和 240px 宽度下都保持 44px 触控区域并留在视口内,且无横向溢出。

证据(前后对比)

改动前:issue 的第一张截图显示 ↑ / ↓ 占据编辑操作行的行首位置。改动后:使用仓库 visuals 夹具本地截图(playwright test --config playwright.visuals.config.ts -g "mobile composer controls",Pixel 7、390x844)并人工查看——编辑操作行只渲染收起键盘和展开两个控件,由行内 spacer 右对齐;在配置了 Add 入口时行首为空;390px 与 240px 下均无残留空隙、错位或横向溢出。截图产物位于被 git 忽略的 client/e2e/visuals/output/screenshots/ 目录,未在此附上。

测试环境

系统 状态
🍏 macOS ✅
🪟 Windows ⚠️
🐧 Linux ⚠️

运行环境

由 Playwright 驱动 packages/web-shell 的 Vite 开发服务器,chromium 通过 playwright install chromium 本地安装;mobile-chromium 项目提供 Pixel 7 触摸模拟。

风险与范围

  • 主要风险或取舍:移动端失去逐条翻阅历史的能力,历史搜索面板成为触屏唯一的回看入口。缓解因素是该面板能力更强(完整列表加过滤)且已适配触屏。
  • 未验证 / 超出范围:iOS/Android 真机键盘行为(仅浏览器模拟,见设计文档待定问题);mobile-webkit Playwright 项目本地未运行(未安装 WebKit 浏览器),由 CI 覆盖。桌面编辑、daemon 路由与宿主工具栏开关均未触及。
  • 破坏性变更 / 迁移说明:对宿主无影响。未改动任何公开 API 或自定义面;被删除的回调仅存在于 web-shell 客户端内部。

设计文档已同步更新中英文版本:English | 简体中文。

关联 Issue

Closes #14137

The touch composer's editing row spent its two leading slots on up/down
buttons that walk the prompt history one entry at a time — a low-frequency
action on the screen where focus matters most. History recall on touch now
goes exclusively through the history search panel, which lists and filters
past inputs and is already touch-ready; the labelled history button (hosts
without the Add entry) and the Add drawer's history entry both remain.

navigatePrevHistory/navigateNextHistory lose their only callers with the
buttons. Desktop arrow keys never used them — the CodeMirror ArrowUp/
ArrowDown keymap handlers implement desktop recall inline — so the two
callbacks and their touch-textarea branches are removed outright rather
than left as dead exports.
@github-actions github-actions Bot added the review/self-reported The linked issue was opened by the PR author (self-reported) label Oct 14, 2026
@pomelo-nwu

Copy link
Copy Markdown
Collaborator Author

E2E test report (mobile composer, touch emulation)

Ran locally on macOS (Apple Silicon) against this branch's Vite dev server, with chromium installed via playwright install chromium.

Mobile composer spec (mobile-chromium project, Pixel 7 touch emulation)

cd packages/web-shell
npx playwright test --config playwright.config.ts --project=mobile-chromium client/e2e/web-shell.composer.mobile.spec.ts

Result: 21 passed (48.3s), exit 0. Notable cases:

  • editing row keeps 44px touch targets and fits the viewport at 390px @smoke and the 240px variant — the rewritten test that now walks every button inside [data-web-shell-mobile-editing-actions] instead of the removed ↑/↓ pair; asserts ≥44px targets, in-viewport placement, and document.documentElement.scrollWidth === viewport width.
  • mobile workspace row stays clear of the editing row @smoke — now anchors on the editing-row container instead of the removed "Previous input" button.
  • mobile history search restores results and draft focus @smoke — confirms the surviving touch history path (search panel restore + focus return, no submit).
  • mobile editing preserves draft, selection and keyboard dismissal @smoke — unchanged behavior of the remaining row controls.

The recurring [vite] http proxy error: /workspace/models ... connect ECONNREFUSED 127.0.0.1:4170 lines in the log are pre-existing dev-server proxy noise (each scenario installs its own mock daemon); every test passed.

Visuals capture (before/after evidence)

npx playwright test --config playwright.visuals.config.ts -g "mobile composer controls"

Result: 2 passed (light + dark). Captured mobile-composer-{light,dark}.png at 390x844 and reviewed them: the editing row renders only the keyboard-dismiss and expand controls, right-aligned by the row spacer; with the Add entry configured the leading slots are empty; no leftover gap, misalignment, or horizontal overflow. PNGs live in the git-ignored client/e2e/visuals/output/screenshots/.

Unit / static gates

  • npx vitest run --config vitest.config.ts — 393 files / 10396 tests passed, exit 0.
  • npx tsc -p tsconfig.json --noEmit — clean.
  • eslint + prettier --check on all changed files (including both design docs) — clean.

Not run locally

  • mobile-webkit project (iPhone 13): WebKit browser not installed in this environment; CI covers it.
  • Real-device iOS/Android checks: browser emulation only.
  • Root npm run build / root typecheck: blocked by a pre-existing, unrelated packages/cli failure on this checkout (missing ink exports such as ReadonlyFrame); the web-shell package gates above are the relevant ones.

@qwen-code-review-bot

qwen-code-review-bot commented Oct 14, 2026 •

Copy link
Copy Markdown
Collaborator

✅ Qwen Triage finished — view run. See the stage comments in this thread for the result.

✅ Qwen Triage 已完成 —— 查看运行。结果见本线程中的各阶段评论。

@qwen-code-review-bot

qwen-code-review-bot commented Oct 14, 2026 •

Copy link
Copy Markdown
Collaborator

Re-run at @wenshao's request. The direction question that stopped my last pass now has an owner's answer — he approved this exact commit (204bd9c) as the surface owner fifteen seconds before triggering the run — so this time it went all the way through. Gate passes. Code review and verdict are in the next two comments.

Template looks good ✓ — every section filled in, both design docs linked in English and Chinese, and the Tested-on table is honest about what wasn't run.

Problem: not a bug, so there is nothing to reproduce — it is a design judgment, and its two factual premises still hold against the current head. The touch restore path really does drop inline @file chips, and history really is still reachable in both toolbar configurations once the arrows go. What changed since the last pass is that the trade is no longer self-endorsed: the owner cc'd on the escalation has signed off on it.

Direction: resolved. My previous pass held here because #14137 was open with need-discussion, its own triage had concluded "don't build it as proposed — owner decision needed", and this PR rewrites a committed spec (docs/design/web-shell/mobile-composer-actions.md) that used to mandate the arrows. @wenshao's approval of this commit is the answer that pass was waiting for, and the spec revision is part of what he approved. One housekeeping note for whoever merges: #14137 still carries need-discussion, and Closes #14137 will shut it on merge — worth dropping that label so the closure doesn't read as unreviewed.

Size: no core paths touched — everything is packages/web-shell/client/** plus two design docs, single package, so the two-tier core gate doesn't apply. 693 changed lines break down as 170 production (30 added / 140 deleted across ChatEditor.tsx, useComposerCore.ts, i18n.tsx), 485 test, 38 docs. Net-negative production code, very much in the spirit of simplicity-first.

Approach: the removal itself matches what I'd have done, and I re-checked the parts that usually go wrong on a deletion PR at this head. navigatePrevHistory / navigateNextHistory are gone from the interface, the hook body and the return object with no remaining consumers; no helper they used becomes dead (restoreSelectedHistoryMatch, rememberPromptHistoryDraftTags, restorePromptHistoryDraftTags, historyBrowseActiveRef, restoreHistoryEntry, mobileTextRef, saveCurrentDraftRef all keep live callers). Both icon imports are dropped and neither is referenced elsewhere in that file. The two i18n keys come out of EN and ZH together, and a green typecheck means no t() call still points at them. Layout doesn't depend on leading children — mobileEditingSpacer is flex: 1, so the trailing controls stay right-aligned with the leading slots empty.

The one thing I want on the record, @pomelo-nwu, is that the PR has outgrown its own description. It is no longer a pure removal: two later commits (deb525e, 80884d8) fix regressions the removal would otherwise have shipped — touch slash commands becoming unrecallable, and a history recall clobbering the stored session draft. Both were causally required, not drive-by, so I'm not asking for a split. But the description still says "the history search panel and every desktop code path" are unchanged, and at this head that is false: the draft-clobber fix deliberately arms historyBrowseActiveRef on both backends, and the touch panel now lists slash entries. A reviewer trusting the description would miss two intentional behaviour changes. Worth a paragraph before merge. Likewise the "393 files / 10396 tests / 21 passed" numbers are the author's own and predate three more commits — I've treated them as a claim, not evidence, and used CI instead.

Risk: no elevated risk signals — none of the nine changed files match the high-risk paths. The Lint & Static failure I flagged last time is gone: it was the stale base missing scripts/check-flyway-migrations.js, and the two merges from main cleared it. It is green on this commit now.

Moving on to code review. 🔍

中文说明

应 @wenshao 的要求重跑。上一次让我停下的「方向」问题现在有了 owner 的答复——他在触发本次 run 前十五秒批准了同一个 commit(204bd9c)——所以这一次完整走完了流程。准入通过,代码审查与最终判定见后两条评论。

模板完整 ✓ —— 各部分均已填写,中英文设计文档都已附上,Tested-on 表格如实标注了未验证项。

问题: 这不是 bug,没有可复现的东西——它是一个设计判断,而支撑它的两个事实前提在当前 head 上依然成立。触屏恢复路径确实不会重建内联 @file chip;箭头移除后,两种工具栏配置下历史入口也确实都还在。与上次相比真正变化的是:这个取舍不再是自我背书——当初被 cc 的 owner 已经签字认可。

方向: 已解决。上一轮我停在这里,是因为 #14137 处于 open 且带 need-discussion,它自己的 triage 结论是「不要按现在的提案做——需要 owner 决策」,而本 PR 改写的是一份已提交规格(docs/design/web-shell/mobile-composer-actions.md),该文档原本明确要求这两个箭头。@wenshao 对这个 commit 的批准,正是那一轮在等的答案,而规格修订本身就是他批准的内容之一。给合并者的一个小提醒:#14137 仍带着 need-discussion,而 Closes #14137 会在合并时关掉它——建议顺手摘掉该标签,免得这次关闭看起来像未经评审。

规模: 未触及核心路径——改动全部在 packages/web-shell/client/** 加两份设计文档,单包,因此核心的两级门禁不适用。693 行改动拆分为:生产代码 170 行(ChatEditor.tsx、useComposerCore.ts、i18n.tsx 共 +30 / −140)、测试 485 行、文档 38 行。生产代码净减少,非常符合 simplicity-first。

方案: 移除本身与我自己的做法一致,并且我在这个 head 上重新核对了删除类 PR 最容易出问题的几处。navigatePrevHistory / navigateNextHistory 已从接口、hook 实现和返回对象中一并移除,没有任何残留调用者;它们用到的辅助函数也没有变成死代码(restoreSelectedHistoryMatch、rememberPromptHistoryDraftTags、restorePromptHistoryDraftTags、historyBrowseActiveRef、restoreHistoryEntry、mobileTextRef、saveCurrentDraftRef 都仍有活跃调用者)。两个图标 import 已移除,且该文件中没有其他地方引用它们。两个 i18n key 在中英文里同步删除,typecheck 全绿说明没有 t() 调用还指向它们。布局不依赖行首子元素——mobileEditingSpacer 是 flex: 1,行首空出后尾部控件依旧右对齐。

有一点我要明确记下来,@pomelo-nwu:这个 PR 已经超出了它自己的描述。它不再是纯粹的移除——后两个 commit(deb525e、80884d8)修掉了移除本身会带进来的两个回归:触屏上已提交的 slash 命令变得无法回看,以及一次历史回看会覆盖已存储的会话草稿。这两处都是因果上必须的,不是顺手改动,所以我不要求拆分。但描述里仍写着「历史搜索面板与所有桌面端代码路径」保持不变,而在当前 head 上这不成立:草稿覆盖的修复是刻意在两个后端上都置位 historyBrowseActiveRef 的,触屏面板现在也会列出 slash 条目。信任描述的评审者会漏掉这两处有意的行为变化。合并前值得补一段说明。同样,「393 个文件 / 10396 个用例 / 21 个通过」这些数字是作者自报的,而且早于后续三个 commit——我把它们当作声明而非证据,实际以 CI 为准。

风险: 无升级风险信号——九个改动文件均未命中高风险路径。上次我提到的 Lint & Static 失败已经消失:那是分支落后导致缺少 scripts/check-flyway-migrations.js,两次从 main 合并已经解决,本 commit 上它是绿的。

进入代码审查 🔍

— Qwen Code · qwen3.8-max-2026-09-02

Reviewed at 204bd9cba00b9f8e3507579a89001e09c1234468 · re-run with @qwen-code /triage

@qwen-code-review-bot

qwen-code-review-bot commented Oct 14, 2026 •

Copy link
Copy Markdown
Collaborator

🖼️ web-shell visual preview

Rendered against a mock daemon (no real backend): the PR base vs this PR head 204bd9c. Only screenshots that changed are shown (flows below, if any, are head-only) — refreshes on every push.

Screenshots · before / after

collab-squad-no-action-light before/after

mobile-composer-dark before/after

mobile-composer-light before/after

Full-resolution recordings (.webm) are attached to the workflow run.

— Qwen Code · web-shell visuals

@qwen-code-review-bot qwen-code-review-bot 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.

Not explored to full depth (tool budget reached): "agent 5": did not execute packages/web-shell/client/e2e/web-shell.composer.mobile.spec.ts (Playwright browsers and a built bundle are not provisioned in this review wor….

中文说明

未探索到全部深度(达到工具调用预算):"agent 5":did not execute packages/web-shell/client/e2e/web-shell.composer.mobile.spec.ts (Playwright browsers and a built bundle are not provisioned in this review wor…。

— qwen3.8-max via Qwen Code /review (v0.25.1)

@wenshao

wenshao commented Oct 15, 2026

Copy link
Copy Markdown
Collaborator

Maintainer verification — ✅ merge-ready

416/416 scripted assertions passed (browser A/B + unit gates + e2e gate + typecheck mutation set + reference census) across the PR build and a base control build.

  • Head verified: bda5898822e70ea641f5f2e5e774f9aeacdcf6c6
  • Base (merge-base with main): 415925cbdabda26da790c596d7b58b65d0fc3279
  • Diff scope: 8 files, all under packages/web-shell/client + the two design docs; +32/−315.
中文摘要

结论:merge-ready,416/416 断言通过。

  • A/B 浏览器实证(同一份证据 spec 跑在两棵树上,mock daemon + Pixel 7 触摸模拟):base 编辑行左侧渲染出 "Previous input" / "Next input" 两个箭头(390px 与 240px 均是),head 上二者完全消失(全页面普查计数 0),其余控件原样保留,44px 触控区域与无横向溢出在两臂都成立;base 在缺失断言上如预期变红,证明断言本身承重。
  • PR 的关键声明"桌面端从不调用这两个回调"经 base 树全量普查证实:base 中 navigatePrevHistory/navigateNextHistory 的生产调用点只有这两个移动端按钮;head 树中对这两个回调及两个 i18n key 的引用为零。
  • 类型系统钉住移除面:在 head 恢复一个箭头按钮后 tsc --noEmit 立即报 TS2339: Property 'navigatePrevHistory' does not exist(变异被击杀);恢复后检查回绿。
  • 门禁:受影响的三个 vitest 文件 353/353 通过(含桌面端 ArrowUp/ArrowDown 历史翻阅回归文件 102 例);head 上完整移动端 Playwright spec 21/21 通过(与作者声明一致);历史搜索面板作为剩余触屏回看入口在两臂均验证可用(恢复文本、0 次提交)。
  • 未覆盖:mobile-webkit(本地未装 WebKit,CI 覆盖)、真机键盘行为、无 addMenu 配置仅用代码审阅+单元测试覆盖(standalone harness 固定启用 addMenu)。见正文 Not covered。

Central claim and A/B proof

Claim: the mobile composer's editing row no longer renders the ↑/↓ history arrow buttons, while the rest of the row and the history-search recall path are unchanged.

Method: one evidence spec (zz-pr14220-evidence.mobile.spec.ts, byte-identical in both trees) drives the real app against the repo's mock-daemon harness under the mobile-chromium (Pixel 7) project, records an OBSERVED JSON per scenario, and asserts the absence of both arrow buttons, 44px touch targets, and zero horizontal overflow at 390px and 240px. The base run must go red on the absence assertions — that is the load-bearing control.

Cell Base 415925cb Head bda58988
Editing row @390px Previous input (44×44 @33–77), Next input (44×44 @81–125), Hide keyboard, Expand editor Hide keyboard (44×44 @265–309), Expand editor (44×44 @313–357)
Editing row @240px Previous input, Next input, Hide keyboard, Expand editor (wraps to a second row) Hide keyboard (@115–159), Expand editor (@163–207) — no wrap needed
Previous input/Next input buttons anywhere on page 1 / 1 0 / 0
Horizontal overflow (scrollWidth vs viewport) none (both widths) none (both widths)
Absence assertion outcome 🔴 red exactly as predicted 🟢 green
History recall via search panel (Add → Input history) restores older saved input, 0 submissions restores older saved input, 0 submissions

Surviving controls keep identical 44×44 geometry on both arms; at 390px the remaining two buttons sit at pixel-identical positions (265–309 / 313–357). At 240px the base row wrapped because the arrows occupied the leading slots; the head row fits on one line.

Verifying the PR's load-bearing claims

  1. "Desktop never called those callbacks" — base-tree census: the only production call sites of navigatePrevHistory/navigateNextHistory are the two deleted buttons (ChatEditor.tsx:3210/3221 at base). Desktop ArrowUp/ArrowDown history walking is implemented inline in the editor keymap and remains covered by useComposerCore.dom.test.tsx (102 tests green at head). ✅
  2. Zero dangling references at head — repo-wide grep for navigatePrevHistory|navigateNextHistory and composerMobile.previousInput|composerMobile.nextInput at head: zero hits. ✅
  3. The removal is pinned by the type system — mutation: restore one arrow button referencing core.navigatePrevHistory and t('composerMobile.previousInput') at head → tsc --noEmit fails with TS2339: Property 'navigatePrevHistory' does not exist on type 'UseComposerCoreReturn' (exit 2). Restored; typecheck green again. ✅
  4. Reviewer Test Plan walk — arrows absent with the addMenu opt-in enabled (browser A/B above); Add → Input history opens the search panel, filters, restores the selected text, returns focus, and does not submit (evidence test, both arms); desktop behavior unchanged (unit gates below). The no-addMenu configuration (labelled history button + Shell toggle in the row) is not reachable from the standalone harness, which hard-enables addMenu (main.tsx: STANDALONE_COMPOSER_TOOLBAR_ADDITIONS = ['addMenu', 'plan']); that branch is covered instead by the untouched {!showAddMenuAction && …} block (visible in the diff context) and by head unit tests that render the labelled Input history button in the row.

Gates at head

Gate Result
vitest on the three touched/related files (ChatEditor.test.tsx, useComposerCore.mobile.dom.test.tsx, useComposerCore.dom.test.tsx) 353/353 passed (220 + 31 + 102)
Committed mobile e2e spec, --project=mobile-chromium 21/21 passed (matches the author's claim)
tsc -p tsconfig.json --noEmit (web-shell) clean
Mutation set: restore-one-arrow-button killed by tsc (TS2339)
Evidence spec at head 3/3 passed
Evidence spec at base (red control) 2 red on the absence assertions exactly as predicted, 1 green

Observation (non-blocking)

The committed suite does not pin the absence of the arrows behaviorally: the rewritten e2e (44px/overflow) and unit tests would stay green if the buttons returned with fresh strings and callbacks — the removal is pinned by the type system (deleted i18n keys and interface members) and by this round's browser census. For a deletion PR that is an acceptable pin; flagging it so the tradeoff is explicit.

Not covered

  • mobile-webkit (WebKit browser not installed locally; CI covers it).
  • Real-device iOS/Android keyboard behavior (browser emulation only, per the design doc's open questions).
  • The host configuration without addMenu at browser level (harness limitation above); covered by code inspection + unit tests.
  • Full web-shell vitest suite (393 files / 10,396 tests) — run by the author; this round ran the three affected files.
  • Windows/Linux manual runs.

Methodology

macOS arm64, Node 22.23.1, pnpm 11.24.0. Two detached worktrees at the verified head and at the merge-base with freshly fetched main; identical lockfile installed with pnpm install --frozen-lockfile --ignore-scripts. The evidence harness needs dists of @qwen-code/qwen-code-core, @qwen-code/acp-bridge, and @qwen-code/sdk; those three packages are byte-identical between the two commits (empty git diff), so they were built once at head and hardlink-copied into the base tree, with node_modules/@qwen-code/* realpaths asserted to resolve inside each own tree. Browser cells ran the identical evidence spec against the repo's mock-daemon Playwright harness (no real backend); OBSERVED JSON and viewport screenshots were recorded per cell before assertions ran. Mutation discipline: edit → gate → restore → git status clean. Assertion ledger: 28 (head browser) + 6 (base cells: 2 predicted-red + 4 recall) + 353 (unit) + 21 (committed e2e) + 3 (typecheck/mutation set) + 5 (census/integrity checks) = 416 pass, 0 fail.

Evidence captures

Before (base @390px) — ↑/↓ occupy the leading slots of the editing row:

base-editing-row-390px

After (head @390px) — arrows gone; keyboard-dismiss and expand remain, right-aligned:

head-editing-row-390px

Before (base @240px) — the row wraps, pushing Expand to a second line:

base-editing-row-240px

After (head @240px) — single row, no overflow:

head-editing-row-240px

History recall still works (head) — Add → Input history → search panel restores the selected input without submitting:

head-history-search-recall


Local verification run by @wenshao; artifacts (report, OBSERVED JSON, raw logs) retained under tmp/pr14220-verify-20261015-080731/.

@qwen-code-review-bot qwen-code-review-bot 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.

Not explored to full depth (tool budget reached): "agent 4": none — all checks I set out to run completed well inside the tool budget.; "agent reverse-audit (round 4)": the three touched Playwright specs were never executed (no browser/dev-host run available in this review); both rewritten specs and the reworked workspace-row s…; "agent 1a": none — but note I could not execute the Playwright specs or tsc /ESLint in this worktree, so the e2e bounds assertions and the typecheck-clean claim rest on st….

Deferred under the convergence posture (round 2, not a blocker) — recorded, not requested in this round:

  • packages/web-shell/client/hooks/useComposerCore.ts:2674 — [review] The touch-side historyBrowseActiveRef reset and its comment now describe a state that cannot occur on that backend (same root as R2-1)
  • docs/design/web-shell/mobile-composer-actions.md:43 — [review] The doc blames chip loss on one-entry stepping, but the surviving search-panel recall path has the identical plain-text limitation
中文说明

未探索到全部深度(达到工具调用预算):"agent 4":none — all checks I set out to run completed well inside the tool budget.;"agent reverse-audit (round 4)":the three touched Playwright specs were never executed (no browser/dev-host run available in this review); both rewritten specs and the reworked workspace-row s…;"agent 1a":none — but note I could not execute the Playwright specs or tsc /ESLint in this worktree, so the e2e bounds assertions and the typecheck-clean claim rest on st…。

收敛姿态下延后(第 2 轮,非阻断)——已记录,本轮不要求修改:共 2 条(原文未翻译,列表见上方英文部分)。

— qwen3.8-max via Qwen Code /review (v0.25.1)

@pomelo-nwu

Copy link
Copy Markdown
Collaborator Author

@qwen-code /takeover

@qwen-code-dev-bot qwen-code-dev-bot added the autofix/takeover Summon the autofix loop to manage this PR (remove to release; needs triage+) label Oct 15, 2026
@qwen-code-dev-bot

Copy link
Copy Markdown
Collaborator

🤝 Takeover engaged: the autofix loop now manages this PR — it will address new review feedback and resolve base conflicts until the label is removed or the round cap is reached. Remove the autofix/takeover label (or comment @qwen-code /takeover stop) to release.

中文说明

🤝 已接管:autofix 循环现在管理此 PR —— 将持续处理新的评审反馈与 base 冲突,直到移除标签或达到轮次上限。移除 autofix/takeover 标签(或评论 @qwen-code /takeover stop)即可释放。

@qwen-code-dev-bot

qwen-code-dev-bot commented Oct 15, 2026 •

Copy link
Copy Markdown
Collaborator

✅ AutoFix round 4 finished — view run. See this round's report below.

中文说明

✅ AutoFix 第 4 轮已完成 —— 查看运行。本轮报告见下方。

@qwen-code-dev-bot

Copy link
Copy Markdown
Collaborator

🤖 AutoFix updated a stale base — the fix did not pass verification, but this PR was behind main, so it merged current main in via update-branch and will retry on the next scan. A stale base (a dependency or symbol main already changed) can fail the build without being the fix's fault; if it still fails once current, it hands off to a human.

What I found before stopping:
Qwen hit the tool-call loop guard during address-review. A human should take over this feedback batch.

See the Qwen Autofix agent step logs for model/tool output.

中文说明

🤖 AutoFix 更新了一个过期的 base —— 修复未通过验证,但本 PR 落后于 main,因此已通过 update-branch 合入当前 main,并将在下次扫描时重试。过期的 base(main 已改动的依赖或符号)可能让构建失败而并非修复本身的错;若 base 更新后仍然失败,将移交人工处理。

Run log: https://github.com/QwenLM/qwen-code/actions/runs/38748627121


🧠 Handled by Qwen Code · model/模型 kimi-k3 · CLI 0.25.1

…and clobbering drafts

Review on #14220 found two regressions from removing the mobile history
arrows: the touch history panel filtered out slash-prefixed entries (the
filter assumes the desktop slash menu re-offers them, which the textarea
backend lacks), and a panel recall no longer armed the history-browse
flag the deleted arrow walk used to set, so a session or workspace
switch migrated the recalled text over the stored draft.

Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
@qwen-code-dev-bot

Copy link
Copy Markdown
Collaborator

🤖 Addressed the latest review feedback (round 2/100). What changed, and what I pushed back on: · 已处理最新评审反馈(第 2/100 轮)。改动内容与我反驳保留之处如下:

Autofix round summary — PR #14220

Commit: deb525efcf — fix(web-shell): keep touch history recall from hiding slash commands and clobbering drafts (3 files, +111/−23).

Feedback dispositions

rc:4276638055, rc:4278671505 — R1-1 (Critical): touch history panel hides submitted slash commands — FIXED

The claim was verified against the code before editing: getSearchMatches filtered out every /-prefixed entry in prompt mode, and with the arrows deleted the panel is the only touch recall path, so a submitted /review … was unrecallable on touch. The touch textarea has no slash completion menu (the desktop rationale for the filter), so the filter is now backend-aware: getSearchMatches returns unfiltered matches on the touch backend and keeps filtering on desktop, where the slash menu re-offers the commands. Desktop behavior is byte-for-byte unchanged.

Pinned by a new case in useComposerCore.mobile.dom.test.tsx (recalls submitted slash commands through history search): type /theme dark, submit, open history search, assert the entry is listed. The mutation probe (dropping the touch exemption) turns exactly this test red; restoring the fix turns it green.

rc:4278671429 — R2-1 (Critical): touch history recall overwrites the stored draft on session switch — FIXED

Deleting the arrow walk removed the last touch-side writer of historyBrowseActiveRef, which also gates the session/workspace-switch draft migration. The flag is now armed in the !view branch of restoreSelectedHistoryMatch (touch-only; the CodeMirror branch and desktop behavior are untouched, and accepted submits still clear it in commitAccepted). All other read sites of the flag were audited: every one except the debounce save, saveCurrentDraft, and the migration lives inside CodeMirror keymap/update-listener code that never runs on the view-less touch backend.

Pinned by a new case in useComposerCore.mobile.dom.test.tsx (keeps the stored draft when a history recall is followed by a session switch), which also proves the handleMobileChange clear re-arms persistence on the next genuine edit. The mutation probe (removing the arm) turns exactly this test red; restoring it turns it green.

rc:4276638064, rc:4278671622 — R1-2 (Suggestion): smoke test only asserts count() > 0 — PARTIALLY ADDRESSED, threads left open

Implemented: the 390px/240px @smoke geometry test now asserts Expand editor and Hide keyboard each render exactly once before the measurement loop, so a dropped control fails instead of shrinking the iterated set. Not implemented: the second browser pass against a host without addMenu — the standalone dev host hard-enables it (STANDALONE_COMPOSER_TOOLBAR_ADDITIONS in main.tsx), so the no-Add row is unreachable from the e2e harness without extending the host, and Playwright browsers are not provisioned in this environment to verify such a pass. The no-Add configuration stays covered by unit tests (preserves shell and history without add opt-in in ChatEditor.test.tsx) and code inspection, which the maintainer verification on this PR explicitly accepted. Both threads got a reply with this reasoning and stay open.

Deferred items recorded in rv:5527648045 (not requested this round)

  • The touch-side historyBrowseActiveRef reset in handleMobileChange "describing a state that cannot occur" is superseded by the R2-1 fix: the flag can be true on touch again, and the reset is load-bearing (pinned by the new test's edit-after-recall leg).
  • The design-doc wording point (chip loss blamed on one-entry stepping while the panel recall has the same plain-text limitation) remains deferred under the round-2 convergence posture.

Failed checks

All five listed checks are CANCELLED (workflow teardown after superseded runs), not failures; nothing actionable.

In-round self-review

Not armed this round (Self-review: off).

Verification

  • COREPACK_HOME=/tmp/corepack-home npm run build — passed (plain npm run build fails in this sandbox: corepack cannot mkdir ~/.cache/node/corepack; the env var redirects the cache to /tmp).
  • npm run typecheck — passed, exit 0.
  • npm run lint — passed, exit 0.
  • npx tsc -p tsconfig.json --noEmit (packages/web-shell) — clean.
  • vitest run client/hooks/useComposerCore.mobile.dom.test.tsx — 33/33 passed (includes both new tests).
  • vitest run client/hooks/useComposerCore.dom.test.tsx client/components/ChatEditor.test.tsx — 322/322 passed.
  • Full web-shell suite — 440 files ran; results across three identical runs were 1, 0, and 3 failures, all inside main-boot.test.tsx and BranchPickerPopover.test.tsx, which never import the composer; both files pass 152/152 in isolation. Diagnosed as parallel-load flake, not this change.
  • Mutation probes: (1) removing the historyBrowseActiveRef arm → the session-switch test fails, restore → green; (2) removing the touch exemption in getSearchMatches → the slash-recall test fails, restore → green.
  • Not run: Playwright e2e (browsers not provisioned in this sandbox). The spec edit only adds two presence assertions using selectors already exercised by neighboring tests in the same file.
中文说明

Autofix 本轮总结 — PR #14220

提交:deb525efcf —— fix(web-shell): keep touch history recall from hiding slash commands and clobbering drafts(3 个文件,+111/−23)。

反馈处置

rc:4276638055、rc:4278671505 —— R1-1(Critical):触屏历史面板隐藏已提交的斜杠命令 —— 已修复

修改前已对照代码核实该论断:getSearchMatches 在 prompt 模式下过滤掉所有以 / 开头的条目;箭头删除后,面板成为触屏唯一的回看入口,因此已提交的 /review … 在触屏端无法找回。触屏 textarea 没有斜杠补全菜单(这正是桌面端过滤的理由),因此过滤现在感知后端:触屏后端返回未过滤的匹配,桌面端保持过滤(其斜杠菜单会重新提供这些命令)。桌面行为逐字节不变。

由 useComposerCore.mobile.dom.test.tsx 新用例(recalls submitted slash commands through history search)钉住:输入 /theme dark、提交、打开历史搜索、断言条目在列。变异探针(移除触屏豁免)恰好使该用例变红;恢复修复后变绿。

rc:4278671429 —— R2-1(Critical):触屏历史召回在会话切换时覆盖已存草稿 —— 已修复

删除箭头路径移除了触屏端最后一个会置位 historyBrowseActiveRef 的写入点,而该标志同时守卫会话/工作区切换时的草稿迁移。现已在 restoreSelectedHistoryMatch 的 !view 分支置位该标志(仅触屏;CodeMirror 分支与桌面行为不受影响,提交被接受时仍由 commitAccepted 清除)。该标志的其余读取点已逐一审计:除防抖保存、saveCurrentDraft 与迁移外,全部位于 CodeMirror keymap/更新监听器内,在无 view 的触屏后端上不会运行。

由 useComposerCore.mobile.dom.test.tsx 新用例(keeps the stored draft when a history recall is followed by a session switch)钉住,该用例同时证明 handleMobileChange 的清除会在下一次真实编辑后重新开启持久化。变异探针(移除置位行)恰好使该用例变红;恢复后变绿。

rc:4276638064、rc:4278671622 —— R1-2(Suggestion):冒烟测试仅断言 count() > 0 —— 部分处理,线程保持开放

已实现:390px/240px 的 @smoke 几何测试在测量循环之前断言 Expand editor 与 Hide keyboard 各渲染一次,因此控件缺失会直接失败,而不是让被遍历的集合悄悄变小。未实现:针对不含 addMenu 的 host 再跑一遍浏览器测试——独立 dev host 固定启用它(main.tsx 的 STANDALONE_COMPOSER_TOOLBAR_ADDITIONS),不扩展 host 就无法在 e2e 中触达无-Add 行,且本环境未安装 Playwright 浏览器、无法验证该用例。无-Add 配置仍由单元测试(ChatEditor.test.tsx 的 preserves shell and history without add opt-in)与代码审阅覆盖,本 PR 的维护者验证已明确接受这一分工。两个线程均已回复说明理由并保持开放。

rv:5527648045 中记录的延后项(本轮不要求处理)

  • 关于 handleMobileChange 中触屏侧 historyBrowseActiveRef 重置“描述了一个不可能出现的状态”一条,已被 R2-1 修复取代:该标志在触屏端重新可能为真,且该重置是承重的(由新用例的召回后编辑一段钉住)。
  • 设计文档措辞问题(把 chip 丢失归咎于逐条步进,而面板召回存在相同的纯文本限制)按第 2 轮收敛姿态继续延后。

失败的检查

列出的五项检查均为 CANCELLED(被后续运行取代后的工作流拆除),并非失败;无可处理项。

轮内自审

本轮未启用(Self-review: off)。

验证

  • COREPACK_HOME=/tmp/corepack-home npm run build —— 通过(本沙箱中直接 npm run build 会失败:corepack 无法创建 ~/.cache/node/corepack;该环境变量将缓存重定向到 /tmp)。
  • npm run typecheck —— 通过,退出码 0。
  • npm run lint —— 通过,退出码 0。
  • npx tsc -p tsconfig.json --noEmit(packages/web-shell)—— 干净。
  • vitest run client/hooks/useComposerCore.mobile.dom.test.tsx —— 33/33 通过(含两个新用例)。
  • vitest run client/hooks/useComposerCore.dom.test.tsx client/components/ChatEditor.test.tsx —— 322/322 通过。
  • web-shell 全量套件 —— 440 个文件;三次相同代码的运行分别出现 1、0、3 个失败,全部位于 main-boot.test.tsx 与 BranchPickerPopover.test.tsx,这两个文件从不引用 composer;二者单独运行 152/152 通过。判定为并行负载下的抖动,与本次改动无关。
  • 变异探针:(1) 移除 historyBrowseActiveRef 置位 → 会话切换用例失败,恢复 → 变绿;(2) 移除 getSearchMatches 的触屏豁免 → 斜杠召回用例失败,恢复 → 变绿。
  • 未运行:Playwright e2e(本沙箱未安装浏览器)。spec 改动仅新增两条存在性断言,所用选择器已被同文件相邻用例使用过。

Base-conflict check · 基分支冲突检查: no conflict with main. · 与 main 无冲突。

🧵 Resolved all 2 selected review thread(s). · 已关闭全部选中的 2 条评审线程。

Re-review when you have a moment. After round 100 this bot stops and leaves the PR for a human. · 有空请复审;第 100 轮后本 bot 停止并将 PR 交给人工。


🧠 Handled by Qwen Code · model/模型 kimi-k3 · CLI 0.25.1

@qwen-code-review-bot qwen-code-review-bot 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.

Test Plan (not a blocker): 10396 tests passed — this review observed 11840, 41870, 635 passed; 21 passed — this review observed 11840, 41870, 635 passed.

Deferred under the convergence posture (round 3, not a blocker) — recorded, not requested in this round:

  • docs/design/web-shell/mobile-composer-actions.md:43 — [review] The doc blames chip loss on one-entry stepping, but the surviving search-panel recall path has the identical plain-text limitation

Convergence: round 3 posted 5 inline comment(s), 4 of them reported for the first time; the previous round posted 3 (1 new). Findings keep coming back to the same files: packages/web-shell/client/hooks/useComposerCore.ts (findings in round 2; 3 more now). The rate of new findings is not falling. A cluster that keeps producing siblings usually means the fixes are treating instances of a shared root cause — triaging that cause before the next round, or splitting an independent cluster into its own pull request, tends to end the loop faster than fixing them one at a time. Batching the remaining fixes and verifying them before the next push, or dropping this PR's reviews to --severity-floor critical, keeps the loop from re-deriving the same set. (Observation only — nothing was withheld from this review because of this observation.)

中文说明

Test Plan(非阻断):10396 tests passed — this review observed 11840, 41870, 635 passed; 21 passed — this review observed 11840, 41870, 635 passed。

收敛姿态下延后(第 3 轮,非阻断)——已记录,本轮不要求修改:共 1 条(原文未翻译,列表见上方英文部分)。

收敛情况:第 3 轮发布了 5 条行内评论,其中 4 条是首次提出;上一轮发布了 3 条(其中 1 条首次提出)。发现反复回到同一批文件:packages/web-shell/client/hooks/useComposerCore.ts(第 2 轮已出过发现,本轮又有 3 条)。新发现的产出速度没有下降。一个不断再生兄弟发现的簇,通常意味着逐条修复只在处理同一根因的实例——先定位并处理该根因,或把独立的簇拆成单独的 PR,通常比逐条修复更快结束循环。把剩余修复攒成一批、验证后再推送,或将本 PR 的评审降到 --severity-floor critical,可以避免循环反复推导同一组发现。(仅为观察——本轮评审未因此扣留任何内容。)

— qwen3.8-max via Qwen Code /review (v0.25.2)

…r history recall

Address round-3 review on the mobile composer history fix:

- A touch history recall armed the history-browse flag, but only a soft-
  keyboard keystroke cleared it, so prompts composed through programmatic
  surfaces (voice, skill/command inserts, composerInput seeds, a rejected
  shell submit's restore) were never persisted as drafts and were lost
  silently. Clear the flag on each touch-backend write path; the rejected
  restore also schedules the save explicitly since the restore itself
  skipped scheduling while armed.
- Hoist the history-browse marker above the backend branch in
  restoreSelectedHistoryMatch so a panel recall on the CodeMirror backend
  no longer overwrites the previous session's stored draft on switch,
  matching the touch and desktop arrow entrances.
- Pin the desktop side of the history-search slash filter and rework the
  re-arm assertion so it discriminates (the prior placement ran after two
  session switches, which had already reset the flag).
- Narrow the editing-row acceptance bullet in both design docs to the
  controls the e2e actually measures and drop the comment's deferral to
  unit tests, which cannot measure layout.

Each new guard is pinned by a focused test verified red under mutation.
@qwen-code-dev-bot

Copy link
Copy Markdown
Collaborator

🤖 Addressed the latest review feedback (round 3/100). What changed, and what I pushed back on: · 已处理最新评审反馈(第 3/100 轮)。改动内容与我反驳保留之处如下:

Address-review round — PR #14220

All five findings from the round-3 review are resolved in commit 80884d8f0c (fix(web-shell): re-arm draft persistence for programmatic writes after history recall).

Findings and dispositions

1. [rc:4287800036] [Critical] Programmatic writes after a touch recall never re-arm draft persistence — RESOLVED

Reproduced before fixing: the new test re-arms draft persistence when a programmatic insert follows a history recall failed on the pre-round code with expected 'my draft' to be 'older saved input/skill ' — the stored draft was never updated after insertText, exactly the reported loss.

Fix, at exactly the prescribed sites:

  • insertText touch branch: clear historyBrowseActiveRef before writing (only when the inserted text is non-empty).
  • setText touch branch: clear before setMobileText.
  • composerInput seeding effect (!view branch): clear before seeding the text.
  • submitSearchMatch rejected-submit restore: clear after the restore, and schedule the save explicitly. The restore arms the flag, so its internal setMobileText skips scheduling; clearing alone left draftDirty false and the blur flush a no-op — the witness test for this site caught that ordering bug in my first implementation before commit.

Per the finding's constraints, the flag is not cleared inside setMobileText itself, and the arming in restoreSelectedHistoryMatch is kept.

Witnesses: four new mobile tests (insert / setText / composerInput seed / rejected shell submit). Each was verified red under a mutation probe that removed exactly that site's re-arm, then green again after restore.

2. [rc:4287800053] [Suggestion] Panel recall still clobbers the stored draft on the CodeMirror backend — RESOLVED

Reproduced before fixing: the new desktop test keeps the stored draft when a history panel recall is followed by a session switch failed on the pre-round code with expected 'older saved input' to be 'my draft' — the session-switch migration wrote the recalled entry over the stored draft.

Fix: hoisted historyBrowseActiveRef.current = true above the backend branch in restoreSelectedHistoryMatch, so a panel recall counts as history browsing on both backends. Per the finding's constraints, the desktop arrow-key path's own flag management (history.isNavigating() / the return-to-draft clear) is untouched, the marker was not pushed into restoreHistoryEntry, and the recall dispatch stays free of user-event annotations. The comment now states the two-backend invariant instead of a touch-only one.

Witness: the new desktop test fails when the hoist is reverted to the touch-only arm (mutation probe), while the existing mobile recall/switch test stays green.

3. [rc:4287800066] [Suggestion] Desktop branch of the slash filter unwitnessed — RESOLVED (test-only)

Added keeps submitted slash commands out of desktop history search matches to the desktop suite (no touch mock): seeds ['/theme dark', 'plain prompt'], opens history search, asserts the matches equal ['plain prompt']. Mutation probe: collapsing the ternary to return matches; turns this test red while the touch-side pin stays green.

4. [rc:4287800084] [Suggestion] Re-arm assertion did not discriminate — RESOLVED (test restructure)

The block ran after two session switches, which reset the flag themselves, so it passed whether or not handleMobileChange cleared the flag. Moved it into a standalone case — re-arms draft persistence when a genuine edit follows a history recall — that recalls, closes search first (so searchMode does not mask the flag), then types and blurs with no intervening switch. Mutation probe: deleting the handleMobileChange clear turns this test red — the exact mutation the old placement survived. The assertion moved within the same file (net-zero removal); the session-switch test keeps its draft-preservation pins.

5. [rc:4287800349] [Suggestion] Geometry test gap vs design-doc promise — RESOLVED (doc-narrowing route)

Took the finding's doc-narrowing route rather than adding a second e2e host scenario, which would grow the diff and require a standalone-host configuration that does not exist. Both design docs now promise only what the e2e measures: the always-rendered controls (Expand editor, and Hide keyboard while the textarea is focused) keep a 44px target and stay inside the viewport at 390px and 240px widths. The e2e comment no longer claims the no-Add contextual buttons are "covered by unit tests"; it states they never render in the standalone host and are not measured by the loop. The existing toHaveCount(1) pins and the per-button height/viewport assertions are unchanged; the spec edit is comment-only.

Notes

  • No findings were declined, deferred, or escalated this round. The "Deferred under the convergence posture" item (design doc line 43) is explicitly recorded-not-requested this round and was left untouched.
  • On the round-3 "Test Plan (not a blocker)" observation about counts: this round's full web-shell run reports 11847 tests — the review-observed 11840 plus the 7 tests this round adds (5 mobile, 2 desktop).
  • Environment note: npm run build initially failed on this runner because corepack could not create its cache under the sandboxed home directory (EACCES: permission denied, mkdir '/home/github-runner/.cache/node/corepack/v1'); re-running with COREPACK_HOME=/tmp/corepack-cache (a writable temp directory) succeeded. Building the workspace prerequisites also required @qwen-code/acp-bridge before @qwen-code/sdk, which the full root build handles in order.

Verification

  • npm run build (run as COREPACK_HOME=/tmp/corepack-cache npm run build — see the environment note above) — passed
  • npm run typecheck — passed
  • npm run lint — passed
  • npx vitest run client/hooks/useComposerCore.mobile.dom.test.tsx (packages/web-shell) — 38 passed
  • npx vitest run client/hooks/useComposerCore.dom.test.tsx (packages/web-shell) — 104 passed
  • npx vitest run client/hooks/useComposerCore.mobile.dom.test.tsx client/hooks/useComposerCore.dom.test.tsx client/hooks/useComposerCore.test.ts (packages/web-shell) — 150 passed
  • npx vitest run --config vitest.config.ts (packages/web-shell, full package suite) — 440 files / 11847 tests passed
  • Pre-fix reproduction, both red on the pre-round code: the programmatic-insert case failed with expected 'my draft' to be 'older saved input/skill ', and the desktop panel-recall case failed with expected 'older saved input' to be 'my draft'
  • Mutation probes (each applied alone, focused test run, then the source restored byte-identical): hoist reverted → desktop panel-recall test red; insertText clear removed → insert test red; setText clear removed → setText test red; composerInput seed clear removed → seed test red; rejected-restore re-arm removed → rejected-shell test red; handleMobileChange clear removed → genuine-edit test red; slash filter collapsed to return matches; → desktop slash-filter test red
  • npm run generate:settings-schema — not applicable (no settings source changed; the build's schema regeneration produced no drift)
  • Integration tests — not applicable (the touched behavior is exercised by the web-shell vitest suites, not the bundled CLI)
  • Playwright e2e — not run locally (the spec edit is comment-only; no test mechanics changed)
中文说明

评审处理轮次 —— PR #14220

第 3 轮评审的全部 5 条发现已在提交 80884d8f0c(fix(web-shell): re-arm draft persistence for programmatic writes after history recall)中解决。

发现与处置

1. [rc:4287800036] [Critical] 触屏回看后的程序化写入不会重新开启草稿持久化 —— 已解决

修复前已复现:新测试 re-arms draft persistence when a programmatic insert follows a history recall 在本轮之前的代码上失败,报错为 expected 'my draft' to be 'older saved input/skill '——insertText 之后存档草稿从未更新,正是报告中的丢失问题。

修复严格落在评审指定的四处:

  • insertText 触屏分支:写入前清除 historyBrowseActiveRef(仅当插入文本非空时)。
  • setText 触屏分支:在 setMobileText 之前清除。
  • composerInput 播种副作用(!view 分支):播种文本前清除。
  • submitSearchMatch 提交被拒后的恢复:在恢复之后清除,并且显式调度一次保存。恢复本身会置位该标志,因此其内部的 setMobileText 会跳过调度;仅清除标志会让 draftDirty 保持为 false、blur flush 成为空操作——我首次实现时的这个顺序问题被该站点的见证测试在提交前抓到。

按该发现的约束,未在 setMobileText 内部清除标志,并保留了 restoreSelectedHistoryMatch 中的置位。

见证:四个新的移动端测试(insert / setText / composerInput 播种 / 被拒的 shell 提交)。每个都通过变异探测验证过:恰好移除该站点的重新开启逻辑时测试变红,恢复后再转绿。

2. [rc:4287800053] [Suggestion] CodeMirror 后端上面板回看仍会覆盖已存草稿 —— 已解决

修复前已复现:新的桌面测试 keeps the stored draft when a history panel recall is followed by a session switch 在本轮之前的代码上失败,报错为 expected 'older saved input' to be 'my draft'——会话切换迁移用召回的条目覆盖了已存草稿。

修复:把 historyBrowseActiveRef.current = true 上提到 restoreSelectedHistoryMatch 的后端分支之前,使面板回看在两个后端上都算作历史浏览。按该发现的约束,桌面方向键路径自身的标志管理(history.isNavigating() 以及回到草稿时的清除)未改动,标志也没有放进 restoreHistoryEntry,且召回的 dispatch 保持不带 user-event 注解。注释现在陈述的是两个后端都成立的不变量,而不是仅限触屏的版本。

见证:把上提退回为仅限触屏置位(变异探测)时新桌面测试变红,而既有的移动端"回看+切换"测试保持绿色。

3. [rc:4287800066] [Suggestion] 斜杠过滤的桌面分支没有见证 —— 已解决(仅测试)

在桌面套件中新增 keeps submitted slash commands out of desktop history search matches(不做触屏模拟):以 ['/theme dark', 'plain prompt'] 预置历史,打开历史搜索,断言匹配结果为 ['plain prompt']。变异探测:把三元表达式塌缩为 return matches; 会让该测试变红,而触屏一侧的固定点保持绿色。

4. [rc:4287800084] [Suggestion] 重新开启的断言不具备区分能力 —— 已解决(测试重构)

该代码块运行在两次会话切换之后,而切换本身就会重置该标志,因此无论 handleMobileChange 是否清除标志断言都会通过。已把它移入独立用例——re-arms draft persistence when a genuine edit follows a history recall——先回看、先关闭搜索(避免 searchMode 掩盖标志),然后在中间不发生会话切换的情况下输入并 blur。变异探测:删除 handleMobileChange 中的清除逻辑会让该测试变红——这正是旧位置能够存活的那个变异。断言在同一文件内移动(净移除为零);会话切换测试保留其草稿保护固定点。

5. [rc:4287800349] [Suggestion] 几何测试缺口与设计文档承诺不符 —— 已解决(采用收窄文档路线)

采用该发现给出的收窄文档路线,而不是新增第二个 e2e 宿主场景——后者会扩大 diff,且需要一个并不存在的 standalone 宿主配置。两份设计文档现在只承诺 e2e 实际度量的范围:始终渲染的控件(展开编辑器,以及输入框聚焦时的收起键盘)在 390px 与 240px 宽度下保持 44px 触控区域并留在视口内。e2e 注释不再声称无 Add 入口的上下文按钮"改由单元测试覆盖",而是如实说明它们在 standalone 宿主中从不渲染、也不被该循环度量。既有的 toHaveCount(1) 固定点与逐按钮的高度/视口断言保持不变;对 spec 的改动仅限注释。

说明

  • 本轮没有拒绝、延后或升级给维护者的发现。"收敛姿态下延后"的条目(设计文档第 43 行)被明确标记为本轮已记录、不要求修改,因此未触碰。
  • 关于第 3 轮"Test Plan(非阻断)"对测试数量的观察:本轮完整 web-shell 运行为 11847 个测试——评审观察到的 11840 加上本轮新增的 7 个测试(移动端 5 个、桌面端 2 个)。
  • 环境说明:npm run build 最初在本 runner 上失败,原因是 corepack 无法在沙箱化的 home 目录下创建缓存(EACCES: permission denied, mkdir '/home/github-runner/.cache/node/corepack/v1');改用可写临时目录(COREPACK_HOME=/tmp/corepack-cache)后成功。构建工作区前置依赖还要求先构建 @qwen-code/acp-bridge 再构建 @qwen-code/sdk,完整的根构建会按此顺序处理。

验证

  • npm run build(以 COREPACK_HOME=/tmp/corepack-cache npm run build 运行——见上方环境说明)——通过
  • npm run typecheck —— 通过
  • npm run lint —— 通过
  • npx vitest run client/hooks/useComposerCore.mobile.dom.test.tsx(packages/web-shell)——38 个通过
  • npx vitest run client/hooks/useComposerCore.dom.test.tsx(packages/web-shell)——104 个通过
  • npx vitest run client/hooks/useComposerCore.mobile.dom.test.tsx client/hooks/useComposerCore.dom.test.tsx client/hooks/useComposerCore.test.ts(packages/web-shell)——150 个通过
  • npx vitest run --config vitest.config.ts(packages/web-shell,包级完整套件)——440 个文件 / 11847 个测试通过
  • 修复前复现,在本轮之前的代码上均为红:程序化插入用例失败于 expected 'my draft' to be 'older saved input/skill ';桌面面板回看用例失败于 expected 'older saved input' to be 'my draft'
  • 变异探测(每次单独施加、运行定向测试、随后将源码逐字节恢复):回退上提 → 桌面面板回看测试变红;移除 insertText 清除 → 插入测试变红;移除 setText 清除 → setText 测试变红;移除 composerInput 播种清除 → 播种测试变红;移除被拒恢复的重新开启 → 被拒 shell 提交测试变红;移除 handleMobileChange 清除 → 真实编辑测试变红;斜杠过滤塌缩为 return matches; → 桌面斜杠过滤测试变红
  • npm run generate:settings-schema —— 不适用(未改动设置源文件;构建中的模式再生成未产生漂移)
  • 集成测试 —— 不适用(本次触动的行为由 web-shell vitest 套件覆盖,而非打包后的 CLI)
  • Playwright e2e —— 未在本地运行(对 spec 的改动仅限注释;未改动任何测试机制)

Base-conflict check · 基分支冲突检查: no conflict with main. · 与 main 无冲突。

🧵 Resolved all 5 selected review thread(s). · 已关闭全部选中的 5 条评审线程。

Re-review when you have a moment. After round 100 this bot stops and leaves the PR for a human. · 有空请复审;第 100 轮后本 bot 停止并将 PR 交给人工。


🧠 Handled by Qwen Code · model/模型 kimi-k3 · CLI 0.25.2

@qwen-code-dev-bot

Copy link
Copy Markdown
Collaborator

🔀 Base updated: red check(s) [Lint & Static (ubuntu-latest, Node 22.x)] pass on current main — merged current main via update-branch; CI will re-run.

中文说明

🔀 已更新 base:红色检查 [Lint & Static (ubuntu-latest, Node 22.x)] 在当前 main 上通过 —— 已通过 update-branch 合入当前 main,CI 将重新运行。

@pomelo-nwu

Copy link
Copy Markdown
Collaborator Author

@qwen-code
image

@qwen-code-dev-bot

Copy link
Copy Markdown
Collaborator

🤖 Reviewed the latest feedback — no changes needed. Why, point by point: · 已审阅最新反馈——无需改动。逐点说明原因如下:

AutoFix round — no code change (clarification needed)

Feedback this round

One new item since the last evaluation: an issue-level comment from @pomelo-nwu
(ic:6204631053) containing only a screenshot, with no accompanying text. The
screenshot shows the branded product's mobile composer during a running turn:
the host's processing placeholder ("Follow up"), and a toolbar of
+ / yolo-mode shield / context-usage ring / a host-rendered gauge / mic /
stop button. No editing row (Expand editor / Hide keyboard) is visible.

CI on head 204bd9c is green and no other feedback is pending.

What I verified against this PR's code

  • On this PR head, the mobile editing row still renders unconditionally
    whenever the touch composer is active (core.mobileComposer && !mobileVoiceActive): at minimum the Expand editor button, plus Hide
    keyboard while the textarea is focused, plus the contextual History/Shell
    controls. Nothing in this PR's diff can remove the row — the diff only
    deleted the two history-arrow buttons inside it, and no CSS changed.
  • The workflow's own visual preview, rendered from this exact head
    (204bd9c, run 38819248866), shows the editing row present in the "after"
    mobile-composer screenshots.
  • The only state that hides the row (live voice active) also hides the
    context-usage ring via data-hide-during-mobile-voice, and the ring is
    visible in the screenshot — so the screenshot is not that state.

Why I did not change code

The screenshot supports two readings, and both exceed what I may decide
autonomously:

  1. Design direction — "make the mobile composer this clean; drop the
    editing row entirely."
    That contradicts this PR's own design doc
    (docs/design/web-shell/mobile-composer-actions.md, both languages),
    whose acceptance criteria explicitly keep the editing row's
    always-rendered controls (Expand editor, and Hide keyboard while focused)
    at 44px targets. Reversing that is a product call for a maintainer, not a
    bot — and it would also delete the Expand/Hide-keyboard features and their
    tests, which a bare screenshot does not clearly request.
  2. Bug report — "the editing row is gone in my build." I could not
    reproduce this from the PR build: per the checks above, the row renders on
    this head. The screenshot comes from the branded host product (host
    placeholder and host toolbar gauge), so if its build genuinely lacks the
    row, the cause is most plausibly a host build that predates feat(web-shell): simplify mobile composer actions #12252 or a
    host-side override — outside this PR's footprint.

Question for @pomelo-nwu

Which reading did you intend?

  • If (1): confirm you want the whole editing row removed (Expand editor +
    Hide keyboard + contextual slots), and I will remove the row, its tests,
    and the design-doc section in the next round.
  • If (2): which build/device produced the screenshot? A commit SHA or build
    info for that deployment would let me chase the missing row concretely.

Verification

  • No code changed this round, so no local build/test commands were run.
  • CI on head 204bd9c: Test, Lint & Static, Integration Tests (no-AK),
    web-shell E2E Smoke, and Capture web-shell visuals all green.
中文说明

AutoFix 本轮——无代码改动(需要澄清)

本轮反馈

自上次评估以来只有一条新反馈:@pomelo-nwu 的一条 issue 级评论
(ic:6204631053),内容只有一张截图、没有任何文字。截图显示品牌产品
在生成运行中的移动端输入框:宿主自定义的 processing 占位文本
("Follow up"),工具栏依次为 + / yolo 模式盾牌图标 / 上下文用量
圆环 / 宿主渲染的仪表盘图标 / 麦克风 / 停止按钮。截图中看不到编辑行
(展开编辑器 / 收起键盘)。

当前 head 204bd9c 的 CI 全绿,没有其他待处理反馈。

我针对本 PR 代码核实的内容

  • 在本 PR head 上,只要触控输入框处于激活状态,移动编辑行仍会无条件
    渲染(core.mobileComposer && !mobileVoiceActive):至少包含"展开
    编辑器"按钮,输入框聚焦时还有"收起键盘",以及按需出现的历史/Shell
    上下文控件。本 PR 的 diff 不可能移除这一整行——它只是删除了行内的
    两个历史箭头按钮,且没有改动任何 CSS。
  • 工作流自己生成的视觉预览(基于当前 head 204bd9c、运行
    38819248866)显示"修改后"的移动端输入框截图中编辑行仍然存在。
  • 唯一能隐藏该行的状态(实时语音激活)还会通过
    data-hide-during-mobile-voice 同时隐藏上下文用量圆环,而截图中圆环
    是可见的——因此截图并非该状态。

为什么我没有改动代码

这张截图有两种解读,而这两种都超出了我可以自主决定的范围:

  1. 设计方向——"让移动端输入框保持这样干净,把整个编辑行去掉。"
    这与本 PR 自己的设计文档
    (docs/design/web-shell/mobile-composer-actions.md,中英两版)
    相矛盾:文档的验收标准明确要求保留编辑行中始终渲染的控件(展开
    编辑器,以及聚焦时的收起键盘),并保证 44px 触控目标。推翻这一点
    是维护者的产品决策,不应由机器人代为决定——而且这会连带删除
    "展开编辑器/收起键盘"功能及其测试,一张没有文字的截图并不能明确
    表达这个要求。
  2. 缺陷报告——"我的构建里编辑行不见了。" 我无法从 PR 构建中复现
    这一问题:根据上面的核查,该行在当前 head 上会渲染。截图来自品牌
    宿主产品(宿主占位文本与宿主工具栏仪表盘图标),因此如果该构建
    确实缺少编辑行,最可能的原因是宿主构建早于 feat(web-shell): simplify mobile composer actions #12252 或存在宿主侧
    的覆盖——这超出了本 PR 的范围。

请 @pomelo-nwu 澄清

您的意图是哪一种?

  • 如果是 (1):请确认要整体移除编辑行(展开编辑器 + 收起键盘 +
    上下文槽位),我将在下一轮移除该行、相关测试以及设计文档章节。
  • 如果是 (2):截图来自哪个构建/设备?提供该部署对应的 commit SHA 或
    构建信息,我就能具体追查编辑行缺失的原因。

验证情况

  • 本轮没有代码改动,因此未运行本地构建/测试命令。
  • head 204bd9c 的 CI:Test、Lint & Static、Integration Tests
    (no-AK)、web-shell E2E Smoke、Capture web-shell visuals 全部通过。

Base-conflict check · 基分支冲突检查: no conflict with main. · 与 main 无冲突。


🧠 Handled by Qwen Code · model/模型 kimi-k3 · CLI 0.25.2

@wenshao
wenshao enabled auto-merge (squash) October 16, 2026 01:00
@wenshao

wenshao commented Oct 16, 2026

Copy link
Copy Markdown
Collaborator

@qwen-code /triage

@qwen-code-review-bot

qwen-code-review-bot commented Oct 16, 2026 •

Copy link
Copy Markdown
Collaborator

Sandboxed verification: ❌ not passed — findings reported (agent verdict) - workflow run

Ran the PR in an isolated, token-free container: A/B against the base build, mock-free harness assertions, targeted gates. Advisory evidence for human reviewers — not a review, an approval, or a CI check.

Scripted assertions: 17 passed · 7 failed · 24 total

Flakiness gate: ✅ 3 changed test file(s) x 5 identical rounds, no divergence

中文 — 判定:❌ 不通过 · 报告了发现(agent 判定)

沙箱验证在隔离、无凭证的容器中执行了该 PR 的代码(与 base 构建 A/B 对照、无 mock harness 断言、定向门禁)。仅作为评审证据,不构成评审、批准或 CI 检查。

脚本断言:17 通过 · 7 失败 · 24 总计

抖动门:✅ 3 changed test file(s) x 5 identical rounds, no divergence

Verification report

Verification report — PR #14220

Verdict: findings — assertions pass=17 fail=7 total=24 (see ledger.md, assertions.json). Verified head OID: 204bd9cba00b9f8e3507579a89001e09c1234468 (git rev-parse HEAD^2); base tip c24e99e9cb74ce616e0b3ff0e0e5b582dbf4b617 (HEAD^1). The arrow-button removal itself is real and complete; the PR also bundles two behavioural changes its description does not disclose, one of which introduces a draft-persistence regression on two write paths, and the headline removal is not pinned by any automated test.

中文摘要
  • 结论:findings。断言 pass=17 fail=7 total=24(见 ledger.md)。
  • A/B 结论:移除 ↑/↓ 按钮这一中心改动本身是真实且完整的——base 仍渲染两个按钮(navigatePrevHistory 出现 3 次),head 为 0,且全仓库对已删除回调与 i18n key 的引用为 0。但该移除没有任何自动化测试固定:把两个按钮加回 head(M1 变异,tsc 仍通过)后,362 个单测与 21 个浏览器用例全绿;head 的 e2e spec 跑在仍带按钮的 base 上也是 21/21 全绿。
  • Findings:(1) PR 实际捆绑了三处改动,正文却称"历史搜索面板与所有桌面端代码路径不变"——getSearchMatches 改变了触屏面板对 slash 命令的可见性,restoreSelectedHistoryMatch 现在在两个后端都会置起 history-browse 标志,二者都是可测的行为变化(见 Corrections)。(2) 该标志置起后只有 4 处写路径重新武装持久化,replaceEditorText 与 clear() 被遗漏:A/B 显示 head 在 recall 之后经这两条路径写入的文本不会落盘(stored 停留在 recall 前的草稿,会话往返后用户看到旧草稿复活),而 base 正确落盘——属本 PR 引入的回归;其 UI 可达面有限(唯一 UI 可达形状 S3 在 base 上同样红,属既有问题)。(3) 中心移除无测试固定(M1 存活)。
  • 未覆盖:每个 commit 的单独归因(shallow checkout 仅可达 1/6 个 commit);mobile-webkit 项目;真机键盘行为;对建议修复的实测(仅给出未验证的最小补丁)。

Central claim + A/B

Central claim (as filed): the mobile editing row no longer renders the Previous input / Next input buttons in either toolbar configuration, and the now-callerless callbacks and their strings are deleted.

The removal is genuine and complete (measured):

control base (HEAD^1) head (HEAD^2)
navigatePrevHistory occurrences in useComposerCore.ts 3 0
arrow buttons rendered in editing row yes (aria-labels at ChatEditor.tsx:3207-8) none
live references to removed callbacks / i18n keys repo-wide — 0
internal @qwen-code/* links resolve into the base worktree 7/7 (readlink -f) n/a

But the claim is not load-bearing in the automated suite. The mutation that re-adds both buttons (M1, no-op handlers + the four strings restored, tsc exit 0) survives everything, and the head e2e spec does not even discriminate base code that still renders the arrows:

instrument expectation if the claim were pinned observed
M1 → ChatEditor.test.tsx + both composer dom suites ≥1 red 362 passed (362)
M1 → Playwright mobile-chromium spec ≥1 red 21 passed, exit 0
head spec run against base tree (arrows present) ≥1 red 21 passed, exit 0

Witness: evidence/03-m1-central-claim-unpinned.png. Static census corroborates: zero references to Previous input / Next input / previousInput / nextInput anywhere under packages/web-shell/client, zero snapshot files, and the rewritten spec's row assertion is expect(await buttons.count()).toBeGreaterThan(0) — which re-added buttons satisfy trivially. The one test that did assert absence (keeps the history arrow controls out of the desktop composer) was deleted by this PR.

Secondary claim A (slash visibility): getSearchMatches now returns slash entries on touch (isShellMode || isTouchComposer). Pinned: reverting just that hunk (m2) turns recalls submitted slash commands through history search red, and the desktop counterpart test pins the desktop side.

Secondary claim B (draft persistence re-arm): restoreSelectedHistoryMatch now sets historyBrowseActiveRef.current = true on both backends, with re-arms added at four write sites. Pinned: removing the flag set (m3) turns both keeps the stored draft … tests red; removing the rejected-submit re-arm (m4) turns its test red. Witness: evidence/02-mutation-matrix.png.

Corrections

The PR body states the history search panel and "every desktop code path" are unchanged. Measured, that is inaccurate in two ways (both are behaviour changes, not defects — stated as corrections to the description, not requests to change code):

  1. The panel's contents changed on touch. getSearchMatches previously filtered /-prefixed entries in all non-shell modes; head returns them when isTouchComposer. Touch users now see submitted slash commands in history search (the PR's own new mobile test pins this), which is a panel behaviour change.
  2. Desktop draft persistence after a panel recall changed. Because restoreSelectedHistoryMatch now sets the browse flag on both backends, a desktop panel recall no longer persists the recalled text until the next edit, and a recalled /… entry no longer pops its slash argument menu (refreshSlashMenuForView early-returns while the flag is set). The new desktop test keeps the stored draft when a history panel recall is followed by a session switch pins the first of these.

Findings

F1 — Draft persistence regression: two write paths never re-arm after a panel recall (measured; bounded reachability)

setMobileText schedules a draft save only when historyBrowseActiveRef is clear. This PR makes restoreSelectedHistoryMatch set that flag, then re-arms it at four sites (insertText, setText, composer-input seeding, rejected shell submit) — but not at replaceEditorText (useComposerCore.ts:4410) or clear() (:4221). After a history-panel recall, a write through either leaves the stored draft stale.

A/B on an identical harness (zz-verify-sibling-sweep.dom.test.tsx, run on both arms; witness evidence/01-sibling-sweep-ab-base-vs-head.png):

scenario base head
S0 genuine edit (control) PASS PASS
S1 replaceEditorText('replaced text') PASS stored="replaced text" FAIL stored="my draft"
S1 after a session round-trip (sharpest consequence) visible=replaced text visible=my draft — the stale draft resurrects on screen
S2 clear() PASS (key removed == emptied) FAIL stored="my draft"
S3 accepted shell submit FAIL stored="my draft" FAIL stored="my draft"
S4 closeSearch(true) (control) PASS PASS

Root cause is attributed, not inferred: the m3 cell (flag set removed) makes S1 pass again (stored="replaced text") while killing the PR's own two stored-draft tests — the same line is both the fix and the cause.

Blast radius / reachability (measured). clear() has no UI caller in the shipped shell (the clearText() hits are a different interface), and replaceEditorText's UI-reachable shapes are closeSearch(true) (restores identical text — S4 passes) and the accepted shell-submit path (S3). S3 fails identically on both arms, so it is pre-existing and not attributable to this PR. Net: F1 is a real state-machine inconsistency on internal hook APIs with limited shipped-UI reach today; it becomes user-visible the moment any host or future UI writes new text through replaceEditorText/clear() after a recall. Severity: minor-to-moderate, not blocking.

Reproduce (head): cd packages/web-shell && npx vitest run --config vitest.config.ts client/hooks/zz-verify-sibling-sweep.dom.test.tsx → S1/S2 red; same file on the base worktree → S1/S2 green.

Suggested minimal fix (proposed, NOT measured in this round)

Add the same re-arm the PR applies elsewhere, at the top of both writers:

const replaceEditorText = useCallback((text: string) => {
  if (isTouchComposer) {
    historyBrowseActiveRef.current = false;
    setMobileText(text);
    return;
  }
  …

and, in clear(), clear the flag before setMobileText(''). Pair it with an S1-shaped test so the axis is pinned. Not applied or re-run here; offered for the author.

F2 — The central removal is not pinned by any automated test (measured)

M1 (buttons re-added at head) survives 362 unit tests and 21 browser tests; the head spec passes 21/21 against base code that still renders the arrows. A future regression that re-introduces the buttons — or a mis-merge — would ship green. The PR deleted the only test asserting absence. Severity: suggestion (coverage), but it is the PR's headline claim. Reproduce: apply tmp/pr14220-verify-*/mutate-m1.mjs apply, run the two suites, observe all-green; revert restores the tree (post-revert control returned exactly to baseline counts).

F3 — Pre-existing, not attributable: accepted shell submit leaves a stale draft (measured on both arms)

S3 fails identically at base and head (stored="my draft" after the composer was cleared by an accepted !cmd submit). Reported for completeness so it is not mistaken for a PR regression; fixing it is out of this PR's scope.

Not covered

  • Per-commit attribution. The checkout is shallow: of the 6 commits in the metadata snapshot only the merge commit 204bd9c is reachable (git rev-list HEAD^1..HEAD^2 = 1). All results are against the aggregate HEAD^1..HEAD diff; the two bundled fix commits (deb525e, 80884d8) were verified only through that aggregate.
  • mobile-webkit Playwright project (WebKit not installed in this image) and real-device iOS/Android keyboard behaviour.
  • The suggested fix for F1 was not applied or re-run; it is labelled proposed-only.
  • The full repo-wide gate was not run; only packages/web-shell vitest (440 files / 11847 tests, exit 0) and tsc --noEmit (exit 0).
  • Visuals/screenshot suite (playwright.visuals.config.ts) not run; the row's visual layout was checked only via the spec's 44px/overflow assertions.
  • One re-run of the sibling harness after a harness-expectation fix was served from a stale vitest transform cache; the recorded first-run logs (which already contained both arms' S2 values) are the evidence used, and the ledger preserves the raw actual= strings.

Methodology

Environment: CI verify lane, node:22-bookworm, merge-ref checkout (HEAD merge, HEAD^1 base tip, HEAD^2 PR head), npm ci + npm run build pre-completed at head; Playwright chromium 149 available via PLAYWRIGHT_BROWSERS_PATH. Base control: git worktree add tmp/base-tree HEAD^1 with root + per-package node_modules hardlink-copied (cp -al) so relative @qwen-code/* symlinks resolve into the base tree — asserted by readlink -f (7/7) and by a positive control (base still contains navigatePrevHistory, head does not). Because the PR touches zero files outside packages/web-shell and docs/, the unchanged packages' pre-built dist/ was staged into the base tree (confound-free, stated). Harnesses: the sibling sweep is a jsdom vitest file driving the real useComposerCore (no stubs); the A/B and mutation cells ran the real Playwright mobile-chromium project against a real Vite dev server. Raw logs, harness sources, and the mutation scripts live in this directory; evidence images in evidence/.

Flakiness gate log

e2e suite, out of gate scope: packages/web-shell/client/e2e/web-shell.composer.mobile.spec.ts
rounds=5 files=3 skipped=1
file packages/web-shell/client/components/ChatEditor.test.tsx: (cd packages/web-shell) npx --no-install vitest run ./client/components/ChatEditor.test.tsx
file packages/web-shell/client/hooks/useComposerCore.dom.test.tsx: (cd packages/web-shell) npx --no-install vitest run ./client/hooks/useComposerCore.dom.test.tsx
file packages/web-shell/client/hooks/useComposerCore.mobile.dom.test.tsx: (cd packages/web-shell) npx --no-install vitest run ./client/hooks/useComposerCore.mobile.dom.test.tsx


per-file results (P=pass F=fail I=infra-exit, one letter per run):
  packages/web-shell/client/components/ChatEditor.test.tsx: PPPPP
  packages/web-shell/client/hooks/useComposerCore.dom.test.tsx: PPPPP
  packages/web-shell/client/hooks/useComposerCore.mobile.dom.test.tsx: PPPPP

verdict: pass
summary: 3 changed test file(s) x 5 identical rounds, no divergence

--- per-invocation detail (full copy in the artifact) ---
round 1 · packages/web-shell/client/components/ChatEditor.test.tsx: P (exit 0)
round 1 · packages/web-shell/client/hooks/useComposerCore.dom.test.tsx: P (exit 0)
round 1 · packages/web-shell/client/hooks/useComposerCore.mobile.dom.test.tsx: P (exit 0)
round 2 · packages/web-shell/client/components/ChatEditor.test.tsx: P (exit 0)
round 2 · packages/web-shell/client/hooks/useComposerCore.dom.test.tsx: P (exit 0)
round 2 · packages/web-shell/client/hooks/useComposerCore.mobile.dom.test.tsx: P (exit 0)
round 3 · packages/web-shell/client/components/ChatEditor.test.tsx: P (exit 0)
round 3 · packages/web-shell/client/hooks/useComposerCore.dom.test.tsx: P (exit 0)
round 3 · packages/web-shell/client/hooks/useComposerCore.mobile.dom.test.tsx: P (exit 0)
round 4 · packages/web-shell/client/components/ChatEditor.test.tsx: P (exit 0)
round 4 · packages/web-shell/client/hooks/useComposerCore.dom.test.tsx: P (exit 0)
round 4 · packages/web-shell/client/hooks/useComposerCore.mobile.dom.test.tsx: P (exit 0)
round 5 · packages/web-shell/client/components/ChatEditor.test.tsx: P (exit 0)
round 5 · packages/web-shell/client/hooks/useComposerCore.dom.test.tsx: P (exit 0)
round 5 · packages/web-shell/client/hooks/useComposerCore.mobile.dom.test.tsx: P (exit 0)

Evidence images

01-sibling-sweep-ab-base-vs-head

02-mutation-matrix

03-m1-central-claim-unpinned

Harness scripts and raw logs are in the workflow run artifacts (7-day retention).

— Qwen Code · sandboxed verification

@qwen-code-review-bot

Copy link
Copy Markdown
Collaborator

Code review

I read the production diff first with my own proposal in hand — for a deletion like this my proposal was exactly what the PR does: drop the two buttons, drop the now-callerless callbacks rather than leave dead exports, drop the two i18n keys, and keep the search panel as the touch recall path. The PR matches it, and then goes one step further in a way I agree with: it fixes two regressions the naive deletion would have shipped.

The two follow-on fixes are load-bearing, not scope creep. Deleting navigatePrevHistory removed the last writer that armed historyBrowseActiveRef on the textarea backend, so a touch history recall would have started overwriting the stored session draft on the next session switch; and because the panel filtered out every /-prefixed entry in prompt mode, promoting it to the only touch recall path would have made submitted slash commands unreachable on touch. Both fixes are the minimum that keeps the removal honest.

I verified the draft-clobber fix actually holds on desktop, since that's where it could silently do nothing. restoreSelectedHistoryMatch now arms the flag above the if (!view) branch, so it covers the CodeMirror backend too. The reason that isn't immediately undone is composerUpdateListener: it clears the flag only when a transaction isUserEvent('input') or isUserEvent('delete'), and a programmatic recall dispatch carries neither. So the flag survives the restore, saveCurrentDraft keeps returning early, and the stored draft survives the session switch. The new desktop test pins exactly that, and the mobile equivalent closes the panel before editing — with a comment explaining that an open panel would mask the assertion via searchMode. That was the vacuous-assertion problem from the last round, and it's genuinely fixed rather than deleted.

The four re-arm sites are all correctly scoped. insertText, setText and the seeded-composerInput path clear the flag inside their isTouchComposer branches, so a programmatic write after a recall is persisted again instead of being silently dropped; the rejected shell-submit path clears it and then calls scheduleDraftSaveRef.current() explicitly, which is right — a refused submit leaves the recalled entry as the live draft, so it does need saving. Each of the four has its own test.

Two things worth naming, neither blocking:

  • A desktop side effect the description doesn't mention. openSlashMenu and refreshSlashMenuForView both gate on the same flag, so on desktop a panel recall of e.g. /theme no longer pops the slash argument menu until the user edits. That matches the intent already documented at the gate ("the user is browsing, not composing"), and I think it's the better behaviour — but it is a desktop change, and the PR body says desktop is untouched.
  • A forward reference. restoreSelectedHistoryMatch (~line 2260) reads historyBrowseActiveRef, which is declared ~40 lines later in the same function scope. Correct at runtime — the callback body only runs after render, so the TDZ is never hit — and lint is green, but it's the kind of ordering that bites whoever moves code later. Moving the declaration above the first callback that uses it would remove the trap.

Still open from the last round, as a Suggestion only: the e2e geometry test can't measure the two contextual buttons this change promotes into the leading slots (Input history, the Shell toggle), because the standalone dev host always opts into addMenu so they never render there. The test now pins Expand editor and Hide keyboard with toHaveCount(1) so a dropped control fails instead of silently shrinking the iterated set, and it documents the gap in a comment — that's an honest partial fix. The contextual buttons are covered by unit tests, which can't measure layout. Fine to land as-is; a host fixture without addMenu would close it if anyone cares later.

I also confirmed nothing is orphaned: every helper the deleted callbacks used still has live callers, and no ArrowUpIcon / ArrowDownIcon reference remains in ChatEditor.tsx.

Test evidence

This is an unattended CI run, so I did not build or execute anything from this branch — the evidence below is the PR's own CI on the reviewed commit, read through the API. The author's self-reported counts in the description (393 files / 10396 tests / 21 Playwright) predate three further commits and are attributed to the author, not counted as evidence here.

Every check on 204bd9c has completed: 10 succeeded, 21 skipped, 0 failed, 0 pending. The skips are the bot-orchestration jobs plus the macOS and Windows Test matrix legs and the with-AK integration leg.

Check Conclusion
Test (ubuntu-latest, Node 22.x) success
Lint & Static (ubuntu-latest, Node 22.x) success
Integration Tests (no-AK, No Sandbox) success
web-shell E2E Smoke (ubuntu-latest, Node 22.x) success
Capture web-shell visuals (ubuntu-latest, Node 22.x) success
Classify PR · assign · label · route · Remind on force-push success
Test (macos-latest, Node 22.x) skipped
Test (windows-latest, Node 22.x) skipped
Integration Tests (CLI, No Sandbox) skipped
…and 18 skipped bot-orchestration jobs skipped

Two notes on that signal. Lint & Static was red on the commit I reviewed last time — not the author's code, but a base 504 commits stale that lacked scripts/check-flyway-migrations.js; the two merges from main cleared it and it is green now. And the unit suite only ran on ubuntu, since the macOS and Windows legs are skipped, so the platform spread in the Tested-on table is not backed by CI either.

Sandboxed verification would settle the remaining gap — and is already in flight. @wenshao triggered @qwen-code /verify on this head (run 38832734754, in_progress at the time of writing); its report will land as its own comment. What CI cannot substantiate is the real-device claim: the author tested on macOS with browser touch emulation only, and by their own account real iOS/Android keyboard behaviour is unvalidated. Specifically unproven by anything I can read here is that on a physical phone the editing row renders correctly with its leading slots empty, and that a panel recall restores text and returns focus without popping the virtual keyboard — the code deliberately avoids focus: on that path for exactly this reason, and emulation is the weakest place to trust it. @qwen-code /tmux is also available to the author (who has write access) if anyone wants the surface driven live rather than emulated.

中文说明

代码审查

我先自己写了一版方案再读生产代码 diff——对这种删除类改动,我的方案与 PR 完全一致:移除两个按钮、连带删掉此后无人调用的回调(而不是留下死导出)、删掉两个 i18n key,并保留搜索面板作为触屏回看入口。PR 与我的方案吻合,并且更进一步,而这一步我是认同的:它修掉了朴素删除会带进来的两个回归。

这两个后续修复是承重结构,不是范围蔓延。 删除 navigatePrevHistory 同时删掉了 textarea 后端上最后一个置位 historyBrowseActiveRef 的写入者,于是一次触屏历史回看会在下次切换会话时覆盖已存储的会话草稿;而由于面板在 prompt 模式下会过滤掉所有以 / 开头的条目,把它提升为唯一触屏回看入口会让已提交的 slash 命令在触屏上无法回看。这两处修复都是让「移除」这件事站得住脚的最小必要量。

我验证了草稿覆盖的修复在桌面端确实生效,因为那里最容易出现「看起来修了其实什么都没做」。 restoreSelectedHistoryMatch 现在在 if (!view) 分支之前置位该 flag,因此 CodeMirror 后端也被覆盖。它不会被立刻撤销的原因在 composerUpdateListener:只有当事务 isUserEvent('input') 或 isUserEvent('delete') 时才清除该 flag,而程序化的回看 dispatch 两者都不带。所以 flag 在恢复之后依然保持,saveCurrentDraft 持续提前返回,已存储草稿在会话切换中得以保留。新增的桌面端测试正是钉住了这一点;移动端对应测试则在编辑前先关闭面板——并有注释说明开着的面板会通过 searchMode 掩盖断言。那是上一轮的空断言问题,这次是真正修好了,而不是把断言删掉。

四处重新置位的作用域都是正确的。 insertText、setText 以及 seeded composerInput 路径都在各自的 isTouchComposer 分支内清除 flag,因此回看之后的程序化写入会重新被持久化,而不是被静默丢弃;shell 提交被拒绝的那条路径清除 flag 后显式调用 scheduleDraftSaveRef.current(),这是对的——被拒绝的提交会把回看到的条目留作当前草稿,确实需要保存。这四处各自都有测试覆盖。

有两点值得点出来,都不阻塞:

  • 一处描述未提及的桌面端副作用。 openSlashMenu 与 refreshSlashMenuForView 都以同一个 flag 作为门禁,因此桌面端从面板回看例如 /theme 时,在用户编辑之前不会再弹出 slash 参数菜单。这与该门禁处已有的注释意图一致("用户在浏览,不是在撰写"),我认为这个行为更好——但它确实是桌面端的变化,而 PR 正文说桌面端未受影响。
  • 一处前向引用。 restoreSelectedHistoryMatch(约 2260 行)读取的 historyBrowseActiveRef 在同一函数作用域内约 40 行之后才声明。运行时是正确的——回调体只在 render 之后执行,永远不会碰到 TDZ——lint 也是绿的,但这种顺序容易让后来移动代码的人踩坑。把声明移到第一个使用它的回调之前就能消掉这个陷阱。

上一轮遗留、仅作为 Suggestion: e2e 几何测试无法测量本次改动提升到行首的两个上下文按钮(Input history、Shell 开关),因为独立 dev host 总是启用 addMenu,它们在那个场景下根本不渲染。测试现在用 toHaveCount(1) 钉住了 Expand editor 和 Hide keyboard,因此控件被丢掉会让测试失败,而不是静默缩小被遍历的集合;缺口也在注释中写明了——这是一个诚实的部分修复。上下文按钮由单元测试覆盖,但单元测试测不了布局。就这样合并是可以的;如果以后有人在意,加一个不带 addMenu 的 host fixture 就能补齐。

我也确认了没有产生孤儿代码:被删回调用到的每个辅助函数都仍有活跃调用者,ChatEditor.tsx 中也不再有 ArrowUpIcon / ArrowDownIcon 引用。

测试证据

本次为无人值守的 CI run,因此我没有构建或执行该分支的任何代码——下面的证据来自被审 commit 上 PR 自己的 CI,通过 API 读取。描述中作者自报的数量(393 个文件 / 10396 个用例 / 21 个 Playwright)早于后续三个 commit,这里只作为作者声明,不计入证据。

204bd9c 上的所有 check 均已完成:10 项成功、21 项跳过、0 项失败、0 项进行中。跳过的是机器人编排任务,以及 macOS / Windows 的 Test matrix 分支和带 AK 的集成分支。

关于这个信号有两点说明。上一轮我审的 commit 上 Lint & Static 是红的——不是作者的代码问题,而是分支落后 504 个提交、缺少 scripts/check-flyway-migrations.js;两次从 main 合并已解决,现在是绿的。另外单元测试只在 ubuntu 上跑过,因为 macOS 与 Windows 分支被跳过,所以 Tested-on 表格里的平台覆盖面也没有 CI 支撑。

沙箱验证可以补齐剩下的缺口——而且已经在跑。 @wenshao 在这个 head 上触发了 @qwen-code /verify(run 38832734754,撰写时状态为 in_progress),其报告会作为独立评论发布。CI 无法证实的是真机部分:作者只在 macOS 上用浏览器触摸模拟测过,并且按其自述,真实 iOS/Android 键盘行为未经验证。具体而言,我在这里读到的任何材料都无法证明:在真实手机上编辑操作行在行首空出的情况下渲染正确,以及一次面板回看能恢复文本并交还焦点而不弹出虚拟键盘——代码正是为此在那条路径上刻意不调用 focus:,而模拟环境恰恰是最不该被信任的地方。作者具备写权限,如有人希望真实驱动该界面而非模拟,@qwen-code /tmux 同样可用。

— Qwen Code · qwen3.8-max-2026-09-02

Reviewed at 204bd9cba00b9f8e3507579a89001e09c1234468 · re-run with @qwen-code /triage

@qwen-code-review-bot

Copy link
Copy Markdown
Collaborator

Confidence: 4/5 — the removal and its two regression fixes are correct and better tested than the code they replace; the missing point is that the PR body no longer describes the PR.

Stepping back: the thing that stopped me two days ago was not the code, it was that a contributor was rewriting a committed spec to remove a capability while the issue behind it still said need-discussion and the only voices in the thread were the author's and the triage bot's. That is now settled in the way it should be — the surface owner looked at it and approved this exact commit. So the direction question is closed, and what's left is a code question, and the code holds up.

What changed my read from "clean removal" to "removal plus two real fixes" is worth dwelling on, because it's the part a skim would miss. The obvious version of this PR — delete the buttons, delete the callbacks, update the tests — would have merged and quietly broken two things: a touch history recall would start eating the stored session draft, and submitted slash commands would become unreachable on touch, since the panel that replaces the arrows filters them out. Both were caught in review and both are now fixed at the right layer. That is the gate working, not the PR ballooning.

Going back to my own proposal from the review pass: it was the same shape, and the PR exceeds it. I would not have thought to arm the browse flag on the desktop backend too, but that's correct — the draft-clobber bug existed there identically, and fixing only the touch branch would have left a known bug in place on the strength of a scope argument. The composerUpdateListener only clears on genuine user input events, so the arming survives a programmatic restore, which is what makes the desktop fix real rather than decorative. I checked that specifically because a fix that gets undone one line later is the easiest thing to merge by accident.

If I maintain this in six months I'd thank the author, with one exception: the description says the search panel and every desktop code path are unchanged, and at this head that is simply not true. Two deliberate behaviour changes sit behind that sentence — the desktop draft fix, and recalled slash entries no longer popping the argument menu. A future reader reconciling this PR against the spec would be misled. That's a paragraph of editing, not a code change, and per the re-run rules an undocumented-scope concern is not grounds to hold a PR that is otherwise sound, so I'm noting it rather than blocking on it. Same for the stale self-reported test counts and the e2e measurement gap on the two contextual buttons, which is documented in the test itself.

The honest limit on my confidence: I read this statically and I did not run it — no PR-derived code executes in this lane. CI on this commit is fully green (10 succeeded, 0 failed, 0 pending), including the web-shell E2E smoke and the unit suites, but only on ubuntu, and the real-device keyboard behaviour the author themselves marked unvalidated stays unvalidated by me. The /verify run @wenshao triggered is still in flight and will report separately; if it contradicts anything above, dismiss this approval and I'll re-run.

Approving, pinned to the commit I reviewed.

中文说明

Confidence: 4/5 —— 移除本身与它的两个回归修复都是正确的,而且测试比它们替换掉的代码更扎实;扣掉的一分在于 PR 正文已经不再准确描述这个 PR。

退一步看整体:两天前让我停下的不是代码,而是「一位贡献者在改写一份已提交的规格、移除一项能力,而背后的 issue 仍写着 need-discussion,讨论串里只有作者和 triage 机器人的声音」。现在这件事以应有的方式落定了——该界面的 owner 看过并批准了这个确切的 commit。所以方向问题已经关闭,剩下的是代码问题,而代码站得住。

我的判断从「干净的移除」变成「移除加两个真实修复」,这一点值得多说两句,因为略读会漏掉。这个 PR 最显然的写法——删按钮、删回调、改测试——合并后会悄悄弄坏两件事:一次触屏历史回看会开始吃掉已存储的会话草稿;而已提交的 slash 命令会在触屏上变得无法回看,因为取代箭头的那个面板会把它们过滤掉。两处都在评审中被抓到,并且现在都在正确的层面上修好了。这是门禁在起作用,不是 PR 在膨胀。

回到我在审查阶段自己写的方案:形状是一样的,而这个 PR 超出了它。我不会想到要在桌面后端也置位 browse flag,但那是正确的——同样的草稿覆盖 bug 在桌面端一模一样地存在,只修触屏分支就会以「控制范围」为借口把一个已知 bug 留在原地。composerUpdateListener 只在真实的用户输入事件时清除该 flag,所以置位在程序化恢复之后依然有效,这正是桌面端修复是真实有效而非装饰性的原因。我专门核对了这一点,因为「下一行就被撤销的修复」是最容易被误合并的东西。

如果六个月后由我来维护这段代码,我会感谢作者,但有一个例外:描述里说搜索面板与所有桌面端代码路径都未改变,而在当前 head 上这根本不成立。那句话背后藏着两处有意的行为变化——桌面端草稿修复,以及回看到的 slash 条目不再弹出参数菜单。将来对着规格复核这个 PR 的人会被误导。这是一段文字修订,不是代码改动;按照 re-run 规则,「范围未记录」这类顾虑不足以扣住一个其他方面都健康的 PR,所以我只是点出来,不作为阻塞。作者自报的测试数量已经过时、以及两个上下文按钮在 e2e 中的测量缺口(测试里已写明),同样如此。

对我这份信心诚实的边界:我是静态读的,没有运行它——这条通道里不会执行任何来自 PR 的代码。本 commit 上的 CI 全绿(10 项成功、0 项失败、0 项进行中),包含 web-shell E2E smoke 与单元测试,但只在 ubuntu 上跑过;作者自己标注为未验证的真机键盘行为,我也未能验证。@wenshao 触发的 /verify run 仍在进行,会单独出报告;如果它与上述任何结论相矛盾,请撤销本次批准,我会重跑。

批准,并钉在我实际审查的那个 commit 上。

— Qwen Code · qwen3.8-max-2026-09-02

Reviewed at 204bd9cba00b9f8e3507579a89001e09c1234468 · re-run with @qwen-code /triage

@qwen-code-review-bot qwen-code-review-bot 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.

LGTM, looks ready to ship. ✅

@wenshao
wenshao merged commit 8bd79c9 into main Oct 16, 2026
50 of 51 checks passed
@qwen-code-review-bot

Copy link
Copy Markdown
Collaborator

Released in v0.25.3.

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

Labels

autofix/takeover Summon the autofix loop to manage this PR (remove to release; needs triage+) review/self-reported The linked issue was opened by the PR author (self-reported)

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Web Shell mobile composer: the ↑/↓ history buttons take prime space for a low-frequency action

4 participants