fix(docs): correct DaemonSessionProvider path in architecture diagram - #11099
Conversation
The Mermaid diagram node showed `web-shell/client/daemon/` but the file lives at `web-shell/client/daemon/session/`. Inconsistent with 14-cli-tui-adapter.md in the same directory which already uses the correct path. Closes #11076 (partial — doc path fix only).
qwen-code-ci-bot
left a comment
qwen-code-ci-bot
left a comment
There was a problem hiding this comment.
@yiliang114 The change itself is correct — I verified it against main. packages/web-shell/client/daemon/DaemonSessionProvider.tsx does not exist; the file lives at packages/web-shell/client/daemon/session/DaemonSessionProvider.tsx, and 14-cli-tui-adapter.md in the same folder already uses the session/ subpath. That line in 01-architecture.md is also the only stale copy of the old path left anywhere under docs/, so the fix is both accurate and complete. No concerns about the diff.
What blocks it is the PR description: it doesn't follow .github/pull_request_template.md. Missing:
## What this PR does— the body uses## Summaryinstead.## Reviewer Test Planand its subsections### How to verify,### Evidence (Before & After),### Tested on— the body has a bare## Test planwith two unticked boxes and no OS table. For a docs-only change,N/Aunder Before/After is exactly what the template asks for.## Linked Issues—Relates to #11076currently sits inline under Risk & Scope; the template wants it under its own heading (use a closing keyword only if this PR is meant to close that issue).- The
<details><summary>中文说明</summary>section — the template asks for a full paragraph-by-paragraph Chinese translation, and it's absent.
Could you rework the description against the template? Once it matches, this should go through without further friction — the one-line change is right.
中文说明
改动本身是正确的，我已对照 main 核实：packages/web-shell/client/daemon/DaemonSessionProvider.tsx 并不存在，该文件实际位于 packages/web-shell/client/daemon/session/DaemonSessionProvider.tsx；同一目录下的 14-cli-tui-adapter.md 也已经使用了 session/ 子路径。而且 01-architecture.md 中这一行是 docs/ 下唯一残留的旧路径，所以这个修复既准确又完整。对 diff 本身没有异议。
阻塞点在于 PR 描述没有遵循 .github/pull_request_template.md，缺少以下内容：
## What this PR does—— 正文用的是## Summary。## Reviewer Test Plan及其子章节### How to verify、### Evidence (Before & After)、### Tested on—— 正文只有一个## Test plan、两个未勾选的复选框，也没有操作系统表格。纯文档改动按模板要求在 Before/After 下写N/A即可。## Linked Issues——Relates to #11076目前写在 Risk & Scope 下面，模板要求单独成节（只有在本 PR 需要关闭该 issue 时才使用 closing keyword）。<details><summary>中文说明</summary>章节 —— 模板要求逐段完整翻译英文正文，目前缺失。
麻烦按模板补全描述。补好之后应该就能顺利推进 —— 这一行改动本身是对的。
— Qwen Code · qwen3.8-max-2026-09-02
chiga0
left a comment
chiga0
left a comment
There was a problem hiding this comment.
No blocking findings. Approved.
Tier: Scan — docs-only, 1 line
What I checked
The factual claim: stale path web-shell/client/daemon/DaemonSessionProvider.tsx → web-shell/client/daemon/session/DaemonSessionProvider.tsx
Verified against the repo at head:
packages/web-shell/client/daemon/— directory listing showssession/as a subdirectory;DaemonSessionProvider.tsxis not directly indaemon/. Old path is wrong. ✓packages/web-shell/client/daemon/session/—DaemonSessionProvider.tsxis present. New path is correct. ✓
The one-line diff matches the claim exactly and is the only instance of the stale path remaining under docs/ (per qwen-code-ci-bot's scan, confirmed).
Cross-check
- qwen-code-ci-bot
CHANGES_REQUESTED: the bot confirmed the diff is factually correct; the only block it raised is PR description not following.github/pull_request_template.md(missing## What this PR does,## Reviewer Test Plansubsections,## Linked Issuesheading, and Chinese translation block). That is a process requirement for this project and the author should address it — but it is not a code or content correctness issue. Confirmed.
Scope
Single markdown file, docs-only. No behavioral surface; rungs 2-3 not applicable.
Reviewed with AI assistance.
|
📄 Docs-only change — the automatic review ran at
Reviewed head: 中文说明📄 纯文档变更 —— 自动评审以 |
|
@qwen-code /triage Description reworked against On the 中文说明描述已按 关于自动评审指出的 commit message 中的 |
|
Sandboxed verification: This PR changes documentation/assets only — there is no code to execute, so a sandboxed verification has nothing to verify. 中文 — 判定：
|
|
Thanks for the PR! Template looks good ✓ — every required heading is present and filled in, the bilingual section mirrors the English one, and nothing is hard-wrapped. Problem: observed, not theoretical. I checked it against Direction: aligned. This package map is the first page an integrator reads for the daemon, and a path that resolves to nothing is worse than no path at all. No CHANGELOG signal to cite for a one-line docs correction, and none needed. Size: not applicable — the only changed file is Approach: the scope feels right and I would not change anything about it. One line, the node id Risk: no elevated risk signals — no file here matches the revert-correlated path set. I also took the description up on its optional offer and resolved every other node in that Moving on to code review. 🔍 中文说明感谢贡献！ 模板完整 ✓ —— 所有必需小标题都在且已填写，中文说明与英文逐段对应，也没有硬换行。 问题：是已观测到的问题，不是理论性加固。我对照 方向：对齐。这张包结构图是接入方读 daemon 时的第一页，一个解析不到的路径比没有路径更糟。一行文档修正没有可引用的 CHANGELOG 信号，也不需要。 规模：不适用 —— 唯一改动文件是 方案：范围合理，我不会改动任何地方。一行改动，节点 id 风险：无升级风险信号 —— 本 PR 没有任何文件命中与 revert 相关的路径集合。我也接受了描述中"可选"的那项邀请，把该 进入代码审查 🔍 — Qwen Code · qwen3.8-max-2026-09-02 Reviewed at |
Code reviewI wrote my own proposal before opening the diff, from the title and the motivation alone: find the provider's node in the package map, insert the missing No blocking findings, and no convention violations — a docs-only edit engages none of the code conventions in AGENTS.md. What I did check, since a Mermaid label is still a parsed artifact:
The author's own risk note is fair — nothing in CI validates paths inside Mermaid labels, which is how this went stale silently. That is not a reason to hold the PR: a label-path checker would be considerably more machinery than the one-line fix it protects. TestingStating plainly what evidence this comment carries. This is an unattended CI run, so per the workflow I built and executed nothing from this PR; and there is no user-visible runtime surface to drive, so a real-scenario terminal capture is N/A. The evidence is two kinds, both real:
Not verified: rendered Mermaid output in a browser. GitHub's preview was not fetched; the claim that rendering is unchanged rests on the syntax check above, not on a screenshot. For a one-word label edit inside an already-quoted string, that is proportionate.
中文说明代码审查我在打开 diff 之前，只依据标题和动机写下了自己的方案：在包结构图中找到该 provider 的节点，补上缺失的 无阻塞性问题，也无规范违反——纯文档改动不涉及 AGENTS.md 中的任何代码规范。由于 Mermaid 标签仍然是需要解析的内容，我确实核查了以下几点：
作者自己提出的风险是成立的——CI 不校验 Mermaid 标签内的路径，这正是它悄悄过期的原因。但这不是暂缓本 PR 的理由：一个标签路径校验器所需的机制，远重于它所保护的那一行修正。 测试明确说明本条评论所承载的证据。这是一次无人值守的 CI 运行，因此按流程我没有构建或执行本 PR 的任何内容；同时它没有用户可见的运行时面，所以真实场景的终端截取为不适用。证据有两类，都是真实的：
未验证：浏览器中 Mermaid 的实际渲染结果。没有抓取 GitHub 预览；"渲染不变"这一判断依据的是上面的语法核查，而非截图。对于一个已在引号字符串内的一词标签修改，这样的深度是相称的。 CI 表格见上方标记区域（内容与英文部分相同，不在此重复）。 — Qwen Code · qwen3.8-max-2026-09-02 Reviewed at |
|
Confidence: 5/5 — clean at every stage, and I went looking for scope creep and found the opposite. Stepping back: this is a one-word correction, and the honest question for something this small is whether it is a real defect or code hygiene dressed up as a fix. It is a real defect. I resolved the label against What actually earns the approval is the restraint, not the size. This PR had every opportunity to balloon: there are stale Against my own proposal written before reading the diff, it matches exactly — no simpler path existed. The node id is untouched so the edges still bind, no Mermaid control characters were introduced, and the label now matches its siblings' form. CI is settled and green with zero failures; the skips are path-filtered legs a docs change legitimately does not trigger. The one thing I would not claim to have verified is rendered Mermaid output in a browser — that rests on a syntax check, not a screenshot. For a label edit inside an already-quoted string I think that is the right amount of care, and I would rather name the gap than imply I looked at a picture. Six months from now this reads as a person who noticed a wrong path, fixed exactly that, and explained why the neighbouring wrong-looking paths were deliberately left alone. Approving. 中文说明Confidence: 5/5 —— 每个阶段都干净；我特意去找范围蔓延，找到的是相反的东西。 退一步看：这是一处一个词的修正，对这么小的改动，诚实的问题是它究竟是真实缺陷，还是把代码洁癖包装成了 fix。它是真实缺陷。我把标签对照 真正让这个 PR 值得批准的是克制，而不是体量。它本来完全有机会膨胀： 与我在读 diff 之前写下的方案相比，它完全吻合——不存在更简的路径。节点 id 未动，因此连边仍然绑定；没有引入 Mermaid 控制字符；标签现在与兄弟节点的写法一致。CI 已落定且为绿，零失败；跳过的是纯文档改动本就不会触发的路径过滤分支。 唯一我不会声称已验证的，是浏览器中 Mermaid 的实际渲染结果——那依据的是语法核查，而不是截图。对于一个位于已加引号字符串内的标签修改，我认为这样的谨慎程度是恰当的；与其暗示我看过渲染图，我宁愿把这个缺口点明。 六个月后再看，这读起来像是一个人发现了一处错误路径、恰好只修了它，并解释了为什么旁边那些看起来也错的引用是刻意保留的。批准。 — Qwen Code · qwen3.8-max-2026-09-02 Reviewed at |
qwen-code-ci-bot
left a comment
qwen-code-ci-bot
left a comment
There was a problem hiding this comment.
LGTM, looks ready to ship. ✅
|
Released in v0.23.1. |
3 participants
What this PR does
Corrects one node label in the Mermaid package map in
docs/developers/daemon/01-architecture.md: the daemon session provider is shown one directory too high, so the label gains thesession/segment it has on disk.Docs-only, one line.
Why it's needed
The package map is the first page an integrator reads, and a wrong path sends them to a directory that does not contain the file. The same folder already spells the path correctly —
14-cli-tui-adapter.md:193readspackages/web-shell/client/daemon/session/DaemonSessionProvider.tsx— so the diagram was the odd one out, and a reader who trusts the diagram over the prose ends up in the wrong place.The two other doc regressions raised in the #9812 review are already gone from
mainand are deliberately not touched here:web-ui.mdno longer exists and nothing links to it, and theACPAdapterparagraph at14-cli-tui-adapter.md:147was rewritten to describe the retirement rather than the removedpostMessagepath.Scope note:
docs/design/anddocs/plans/still containpackages/webuipaths, and they stay. Those are dated design records describing the tree as it was when they were written; rewriting their paths would falsify the record. Only living documentation underdocs/developers/is corrected, and a scan ofdocs/developers,docs/usersandREADME.mdfinds no remainingpackages/webuior@qwen-code/webuireference.Reviewer Test Plan
How to verify
git ls-files 'packages/web-shell/client/daemon/**/DaemonSessionProvider.tsx'returnspackages/web-shell/client/daemon/session/DaemonSessionProvider.tsx.git grep -n 'DaemonSessionProvider.tsx' docs/developers/daemon/shows the samedaemon/session/segment in01-architecture.mdand14-cli-tui-adapter.md.flowchart TBblock also resolves onmain—serve/,acp-bridge/src/,core/src/tools/,sdk-typescript/src/daemon/,channels/base/src/DaemonChannelBridge.ts,vscode-ide-companion/src/services/daemonIdeConnection.ts. The subgraph labels give the package and the nodes give the file, so the shorter forms are the diagram's convention, not further staleness.Evidence (Before & After)
N/A — docs-only, no user-visible or runtime surface.
Tested on
Environment (optional)
N/A — no build or test run; the diff contains no executable code. Mermaid renders in GitHub's Markdown preview unchanged apart from the label text.
Risk & Scope
packages/webuipaths indocs/design/anddocs/plans/, left intact on purpose (see above). No behaviour, no code.Linked Issues
Relates to #11076. Follows the doc regressions raised in the #9812 review.
中文说明
这个 PR 做了什么
修正
docs/developers/daemon/01-architecture.md中 Mermaid 包结构图里的一个节点标签：daemon session provider 的路径少了一层，补上磁盘上实际存在的session/。纯文档，一行。
为什么需要
这张包结构图是接入方读到的第一页，路径写错会把人指向一个并不包含该文件的目录。同一个目录下的文档其实已经写对了——
14-cli-tui-adapter.md:193写的是packages/web-shell/client/daemon/session/DaemonSessionProvider.tsx——所以出错的只有这张图；读者若更相信图而不是正文，就会找错地方。#9812 评审中提出的另外两处文档回归在
main上已经不存在，这里刻意不动：web-ui.md已被删除且无任何链接指向它；14-cli-tui-adapter.md:147的ACPAdapter段落已改写为描述该路径的退役，而不是描述已被移除的postMessage机制。范围说明：
docs/design/与docs/plans/中仍存在packages/webui路径，且保持不动。那些是带日期的设计记录，描述的是撰写当时的代码树，改写其路径等于篡改记录。这里只修正docs/developers/下的活文档；对docs/developers、docs/users与README.md的扫描显示已无任何packages/webui或@qwen-code/webui引用。评审者验证计划
如何验证
git ls-files 'packages/web-shell/client/daemon/**/DaemonSessionProvider.tsx'返回packages/web-shell/client/daemon/session/DaemonSessionProvider.tsx。git grep -n 'DaemonSessionProvider.tsx' docs/developers/daemon/显示01-architecture.md与14-cli-tui-adapter.md使用相同的daemon/session/层级。flowchart TB块中其余路径在main上均可解析——serve/、acp-bridge/src/、core/src/tools/、sdk-typescript/src/daemon/、channels/base/src/DaemonChannelBridge.ts、vscode-ide-companion/src/services/daemonIdeConnection.ts。subgraph 标签给出包名、节点给出文件名，因此那些较短的写法是这张图的表达约定，不是新的过期路径。证据（改前 / 改后）
不适用——纯文档，无用户可见行为与运行时面。
测试环境
运行环境（可选）
不适用——未执行构建或测试；diff 中不含可执行代码。除标签文本外，Mermaid 在 GitHub Markdown 预览中的渲染不变。
风险与范围
docs/design/与docs/plans/中的历史packages/webui路径，刻意保留（见上）。无行为改动、无代码改动。关联 Issue
Relates to #11076。承接 #9812 评审中提出的文档回归。