Skip to content

[PR] docs: dsh experimental-inspector fetch-accessor coexistence note (#5148) - #5151

Open
ranxianglei wants to merge 9 commits into
masterfrom
2026-10-29_dsh-inspector-fetch-docs
Open

ranxianglei wants to merge 9 commits into
masterfrom
2026-10-29_dsh-inspector-fetch-docs

Conversation

@ranxianglei

Copy link
Copy Markdown
Owner

Closes the in-scope (no-decision) part of #5148.

What

Documents how bili's dsh native plugin interacts with dsh's
@deepseek-ai/dsh-experimental-inspector (enabled by Developer Tools), which
fails to activate because bili installs globalThis.fetch as a guarded accessor
(the #1158 self-heal) while the inspector hard-throws on accessors.

Adds a coexistence note to the dsh section of CLIENTS.md + CLIENTS.zh-CN.md:

Scope / behavior change

Docs only. No runtime code touched. No new config/env (BILI_RECLAIM_FETCH_PATCH already exists and is documented). Zero regression risk.

Out of scope (tracked separately under #5148, pending owner approval)

The root-cause fix lives in the external dsh project — making the
experimental-inspector degrade only its fetch-capture instead of throwing the
whole entry when it sees an accessor. Cross-repo submission requires owner
sign-off before anything goes upstream.

@github-actions

github-actions Bot commented Oct 29, 2026 •

Copy link
Copy Markdown
Contributor

📦 Built Package Artifact

Branch: 2026-10-29_dsh-inspector-fetch-docs (efd7a2d)

Latest build — install this exact version

npm install -g billion-context@0.0.10677-pr.5151.b41e200

This exact version is immutable and reproducible. The rolling pr tag tracks the newest test build across all PRs.

Recent builds (commit → version)

commit version build
efd7a2d billion-context@0.0.10677-pr.5151.b41e200 build
4097ca4 billion-context@0.0.9138-pr.5151.2399c7b build
8414e01 billion-context@0.0.7388-pr.5151.fcf9280 build
2556c1c billion-context@0.0.7351-pr.5151.4a19129 build
14a1d63 billion-context@0.0.7265-pr.5151.2a4c7f0 build
76fca38 billion-context@0.0.7011-pr.5151.abb349e build
7462fae billion-context@0.0.6786-pr.5151.8920a97 build
f866125 billion-context@0.0.6785-pr.5151.3ae7bbd build

Download artifact

  1. Download the artifact from the Actions run
  2. Extract the tarball and install:
tar xzf billion-context-pr5151.tgz
npm install -g package

Install instructions are refreshed on each push.

@ranxianglei ranxianglei changed the title docs: dsh experimental-inspector fetch-accessor coexistence note (#5148) [PR] docs: dsh experimental-inspector fetch-accessor coexistence note (#5148) Oct 29, 2026
@ranxianglei

Copy link
Copy Markdown
Owner Author

🤖 Powered by ework · qwen3.8-27b

(ai: vllm-qwen/qwen3.8-27b)

查重: 不重复。本条是 GitHub PR #5151 的平台镜像(来源 issue #5148,用户 free5394 报告 dsh experimental-inspector 因 bili 的 fetch 访问器无法激活)。相关条目:#5148(本 PR 的来源 issue)、#3528(更早的同机制报告——同一 @deepseek-ai/dsh-experimental-inspector 组件在 DSH Desktop 场景报过同样的访问器冲突,CONFIGURATION.md 的 BILI_RECLAIM_FETCH_PATCH 条目里已有「共存提示(#3528)」段落)。本 PR 与 #3528 的关系是互补:#3528 写在配置文档、以 BILI_RECLAIM_FETCH_PATCH=0 为主解法;本 PR 写在 dsh 车道客户端文档、按影响从小到大给出三种绕行(首选 captureFetch: false),不冲突。

初审结论(方向关 + 事实核对):docs-only,两个文件纯新增 20 行(CLIENTS.md / CLIENTS.zh-CN.md 各一段),零运行时代码、零新配置。PR 描述里的技术断言我逐条对 master 源码核过:

CI 现状(镜像缓存无数据时单次 GitHub 只读拉取):static-gates(ubuntu)/mysql-contract/gate/lane-guard/docs-sync/plan/docs-coverage 已绿;test 分片(ubuntu+windows × node22/24)仍在跑。

我在做的收尾:分支落后 master 15 个提交,但 master 这 15 个提交未触碰这两个文件(已验证零重叠);我把它 rebase 到最新 master,并补一个一行级小修——新增段落交叉引用 CONFIGURATION.md 既有的 #3528 共存提示(避免两处文档互不知晓)。随后回贴最终八问结论。

@ranxianglei
ranxianglei force-pushed the 2026-10-29_dsh-inspector-fetch-docs branch from f866125 to 7462fae Compare October 29, 2026 13:49
@ranxianglei

Copy link
Copy Markdown
Owner Author

🤖 Powered by ework · qwen3.8-27b

(ai: vllm-qwen/qwen3.8-27b)

收尾完成,可合并。 我做过的动作(均已验证):

  1. rebase 到最新 master——原分支落后 15 个提交,但这 15 个提交未触碰本 PR 的两个文件(零重叠),rebase 干净无冲突;分支不再是 behind。
  2. 一行级小修(独立标注 commit 7462faeb4):新增段落补了指向 CONFIGURATION.md 既有「共存提示([plugin] Guarded globalThis.fetch accessor blocks coexisting fetch observers (dsh experimental-inspector fails to load) #3528)」的交叉引用——同一 inspector 组件更早经 DSH Desktop 场景报过同样冲突并已在配置文档记录,两处文档此前互不知晓。英文/中文各加一句,无其他改动。
  3. push 后新 head = 7462faeb4。CI:镜像缓存当前 30 项检查、0 失败、8 项在跑(test 分片);静态门禁(static-gates/mysql-contract/docs-sync/docs-coverage/lane-guard/gate/plan)已绿。纯 .md 改动,最终绿以远端 CI 为准。

结构化八问:

  1. 是什么问题:bili 的 dsh 原生插件与 dsh Developer Tools 的实验版 inspector 同时启用时,inspector 因读不到 data 属性形态的 fetch 而整条 entry 激活失败(1 entry did not activate),而用户在此之前没有任何文档入口能发现这个交互和绕行办法。
  2. 什么原因导致的:加载顺序 + 机制冲突。dsh profile 先加载 bili、后加载 inspector;bili 默认把 globalThis.fetch 装成受守卫访问器(方式一安装的dsh 原生插件:llm-pi-ai 传输(OpenAI SDK + options.fetch 注入)绕过 globalThis.fetch 拦截,模型请求不进代理、压缩完全不生效 #1158 自愈,src/agent/fetch-chain/index.ts:192-234,getter 返回链头/setter 把第三方安装重链为下游);inspector 的守卫是「描述符无 value 就整条抛错」。根本修法在 inspector 侧(遇访问器只降级 fetch-capture)——跨仓库投递等你拍板,已在 Bug: billion-context fetch accessor prevents dsh experimental-inspector from activating #5148 跟踪,未动。
  3. 影响范围与严重程度:仅「bili dsh 原生插件 + Developer Tools」同时启用的用户;主服务、模型流量、压缩全部正常,只有 inspector 的 fetch 观察失效且有现成绕行 → p2 体验瑕疵。(本 PR 本身零运行时影响。)
  4. 怎么修复的:docs-only。CLIENTS.md / CLIENTS.zh-CN.md 的 dsh 段各加一段:成因(非 bili 单方之错)、为何访问器默认保持开启(关掉会重新引入 方式一安装的dsh 原生插件:llm-pi-ai 传输(OpenAI SDK + options.fetch 注入)绕过 globalThis.fetch 拦截,模型请求不进代理、压缩完全不生效 #1158 静默失路由)、三种绕行按影响从小到大(captureFetch: false /关 Developer Tools/BILI_RECLAIM_FETCH_PATCH=0 并注明代价)、根因修法指向 inspector 侧;外加上述 [plugin] Guarded globalThis.fetch accessor blocks coexisting fetch observers (dsh experimental-inspector fails to load) #3528 交叉引用小修。
  5. 是否有回归问题:无。未触碰任何运行时代码、配置面、wire、持久化;高危模式逐项过——语义漂移/fail-fast 变吞错/阈值·时序·输出格式变化均不涉及。特别确认:BILI_RECLAIM_FETCH_PATCH 默认值(on)原样未动——没有为迁就 inspector 去关默认访问器,那才是真回归。
  6. 是否破坏缓存:不涉及。
  7. 是否改了设计方向:不改。维持 方式一安装的dsh 原生插件:llm-pi-ai 传输(OpenAI SDK + options.fetch 注入)绕过 globalThis.fetch 拦截,模型请求不进代理、压缩完全不生效 #1158 访问器自愈为既定默认行为;把「与第三方 fetch 观察者共存」的协调责任明确划到 inspector 侧。
  8. 是否引入了新的配置:没有。BILI_RECLAIM_FETCH_PATCH 已存在且早已在 CONFIGURATION.md 记录;本次只补它与 inspector 交互的说明。

判定:方向正确、事实全部对源码核过、diff 收敛到目的本身、CI 无红 → 建议合并(合并操作留给你,PR: https://github.com/ranxianglei/billion-context/pull/5151)。跨仓库那条(dsh inspector 降级而非抛错)维持等待你批准,批准后我再走投递流程。

@ranxianglei

Copy link
Copy Markdown
Owner Author

🤖 Powered by ework · qwen3.8-27b

(ai: vllm-qwen/qwen3.8-27b)

第五轮同步完成:已合入最新 master@e5e210865(零冲突;master 未触碰本 PR 的两个文档文件),新 head 8414e012d,base=e5e210865,mergeable=True,state=blocked(等 CI 重跑);相对 base 的改动仍只有 CLIENTS.md / CLIENTS.zh-CN.md 两处说明,无夹带。

Resolve the CLIENTS.md / CLIENTS.zh-CN.md conflict caused by the docs/
reorganization (#5920): master moved both files to docs/clients.md and
docs/clients.zh-CN.md leaving one-line redirect stubs at the old root paths.
Keep master's stubs at the root and land the PR's dsh experimental-inspector
coexistence note (#5148) in the new docs/ home instead. Adapted the note's
cross-reference from CONFIGURATION.md to configuration.md to match the
lowercase relative-link convention used throughout docs/. No other file
changed.
@ranxianglei

Copy link
Copy Markdown
Owner Author

🤖 Powered by ework · qwen3.8-27b

[bot] 🏷
(ai: vllm-qwen/qwen3.8-27b)

Done — merge conflict on PR #5151 resolved and pushed. The PR is now mergeable: true (state went dirty → behind).

What blocked it: master's docs reorganization (#5920) moved CLIENTS.md → docs/clients.md and CLIENTS.zh-CN.md → docs/clients.zh-CN.md (lowercased), leaving one-line redirect stubs at the old root paths. This PR had added its dsh experimental-inspector coexistence note (#5148) into the old root files, so "insert a paragraph at the old path" collided with "master rewrote the whole file into a stub" → modify/rewrite conflict (mergeable_state: dirty).

How I resolved it: merged current master into the branch; kept master's redirect stubs at the two root paths (did not resurrect old content); re-landed the PR's two paragraphs in their new home — docs/clients.md (EN, after the dsh Developer Tools paragraph / before ## Kimi Code) and docs/clients.zh-CN.md (ZH, after the #2837 paragraph / before ## Kimi Code). Verified the net diff vs current master is exactly those two paragraphs and nothing else (byte-identical otherwise). Pushed as a clean fast-forward 8414e01 → 4097ca4dd7 (no force-push).

Content change flagged: the note cross-referenced CONFIGURATION.md; master moved that file to docs/configuration.md too, so I changed the reference to configuration.md (both EN + ZH) to match the lowercase relative-link convention already used throughout docs/clients.md — otherwise it would be a dead link after the move. Old → new: CONFIGURATION.md → configuration.md.

Ops note: a local git remote alias was not usable this run, so I fetched/pushed against the canonical upstream GitHub repository URL directly instead of git push origin. Single clean fast-forward; no force-push.

CI: will re-run on the new head 4097ca4dd7; no check results existed at push time. Docs-only change, so no local suite run needed (CI validates remotely).

结构化八问

  1. 是什么问题:PR [PR] docs: dsh experimental-inspector fetch-accessor coexistence note (#5148) #5151(docs:dsh experimental-inspector fetch-accessor 共存说明,Bug: billion-context fetch accessor prevents dsh experimental-inspector from activating #5148)与 base 分支 master 冲突(mergeable_state: dirty),无法合并。
  2. 什么原因导致的:master 的文档重构([PR] docs: batch-migrate 14 root-level documents into docs/ (consolidation + master index + redirect placeholders) | docs: 14 个根级文档一次性迁入 docs/(集中 + 总目录 + 重定向占位) #5920)把 CLIENTS.md/CLIENTS.zh-CN.md 整体迁到 docs/clients.md/docs/clients.zh-CN.md(小写),原 root 路径只留一行重定向占位;本 PR 的两段内容却加在旧的 root 文件里。于是「整文件被改写成占位」撞上「旧位置插段」,形成 modify/rewrite 冲突。本质是落点随迁移失效,不是内容本身有矛盾。
  3. 影响范围与严重程度:只影响这一个 docs PR 的可合并性;不影响任何运行时行为、代码、缓存或用户面。严重度=该 PR 合并被阻塞(dirty),修后即恢复;无功能/数据/安全影响。
  4. 怎么修复的:把当前 master 合并进 PR 分支;root 两个文件保留 master 的重定向占位(不回填旧内容),把 PR 的两段共存说明落到新家 docs/clients.md(英文)与 docs/clients.zh-CN.md(中文)的对应位置。最终分支 = 当前 master + 仅这两段,其余逐字节一致。
  5. 是否有回归问题:无。合并后 git diff origin/master..HEAD --stat 只有 docs/clients.md(+19) 与 docs/clients.zh-CN.md(+2),其余与 master 完全一致;root 两文件=master 占位。未碰 src//kernel//配置 schema/持久化/wire,无 fail-fast 变吞错、无阈值/时序/输出格式漂移。
  6. 是否破坏缓存:不涉及(纯文档,不触碰缓存/重试/增量游标/去重键)。
  7. 是否改了设计方向:无架构/边界/数据流转向。唯一内容级调整:把两段里的交叉引用 CONFIGURATION.md → configuration.md(小写相对名),因 master 同样把该文件迁到了 docs/configuration.md、且 docs/ 内一律用小写相对链接(如同文件的 technical-notes.md),不改会留下指向不存在路径的坏引用。属「落点迁移」的必然配套,非方向变更。
  8. 是否引入了新的配置:不涉及(无新 env/config/默认值)。

判定:可合入。 冲突已解决并推送(head=4097ca4dd7,干净 fast-forward,未 force-push);mergeable: true,dirty → behind。「behind」仅表示 master 在我合并之后又前进了约 25 个提交——活跃仓库的正常现象,非冲突、不阻塞合入。

@ranxianglei

Copy link
Copy Markdown
Owner Author

🤖 Powered by ework · qwen3.8-27b

[bot] 🏷
Re-verified live just now (this forward repeats the earlier “dirty” detection — it looks like it was queued before my push propagated). PR #5151 is still resolved:

  • head 4097ca4dd7 — unchanged from my push
  • mergeable: true, mergeable_state: behind (not dirty)

No new conflict, so no further action this run. On behind: master keeps advancing (≈25 commits since my merge), so the branch reads behind rather than clean — normal on this active branch and non-blocking (mergeable: true). The full root-cause + 八问 are in my previous comment above; nothing has changed since then.

@ranxianglei

Copy link
Copy Markdown
Owner Author

🤖 Powered by ework · qwen3.8-27b

[bot] 🏷
Re-verified live (this is the 3rd identical “dirty” forward — the live state has been resolved ever since my first fix):

  • Conflict: still resolved — head 4097ca4dd7 unchanged, mergeable: true, mergeable_state: behind (not dirty). Nothing has changed since my push.
  • CI: green — all 30 upstream checks completed, 0 failed, 0 pending (test shards across ubuntu/windows × node 22/24, static-gates, config-surface-guard, lane-guard, gate, lock-sync all pass; the real-client e2e lanes are skipped by their own gates for a docs-only change, as expected).

Net: no conflict + CI green → the PR is merge-ready. These recurring “dirty” forwards look like a stale re-detection on the detector side; there’s nothing left for me to fix on this branch. Full root-cause + 八问 remain in my earlier comments above.

@ranxianglei

Copy link
Copy Markdown
Owner Author

🤖 Powered by ework · qwen3.8-27b

[bot] 🏷
(ai: vllm-qwen/qwen3.8-27b)

Final status on PR #5151 (dsh experimental-inspector fetch-accessor coexistence note):

What was done — Resolved the merge conflict. Root cause: master’s docs reorg (#5920) moved CLIENTS.md/CLIENTS.zh-CN.md → docs/clients*.md (lowercased), leaving redirect stubs at the old root paths; the PR’s added paragraph sat at the old root path → a modify-vs-full-rewrite conflict. I merged master into the branch, kept master’s stubs at the root, and re-applied the two doc paragraphs to their new home (docs/clients.md, docs/clients.zh-CN.md). Pushed 4097ca4dd7 as a clean fast-forward (no force-push).

Verified live this run — head 4097ca4dd7 unchanged; mergeable: true, mergeable_state: behind (not dirty); CI green (30 checks completed, 0 failed, 0 pending).

Outcome — no conflict + green CI → the PR is merge-ready. The repeated “dirty” forwards were stale re-detections; nothing remains for me to fix. Merging is human-only — ready whenever you are.


中文八问:

  1. 是什么问题:该 docs PR 与 base 分支产生合并冲突(merge_state=dirty)。
  2. 什么原因导致:master 文档重组 [PR] docs: batch-migrate 14 root-level documents into docs/ (consolidation + master index + redirect placeholders) | docs: 14 个根级文档一次性迁入 docs/(集中 + 总目录 + 重定向占位) #5920 把 CLIENTS.md/CLIENTS.zh-CN.md 移到 docs/clients*.md(小写),旧根路径只留一行重定向存根;而 PR 的新增段落落在旧根路径上,形成“局部修改 vs 整文件重写”的冲突。
  3. 影响范围与严重程度:仅阻塞该 docs PR 的合并,无任何运行时/功能影响。
  4. 怎么修复的:把 master 合入 PR 分支、保留 master 在旧根路径的存根,再把两段文档重新落到新位置 docs/clients*.md;干净快进推送 4097ca4dd7(未强推)。
  5. 是否有回归问题:无。我这次的提交相对其合并基线(master@33104b1)仅新增两段文档(docs/clients.md +19 / docs/clients.zh-CN.md +2),未触碰任何其他文件;此后 master 又前进了若干提交(故状态为 behind),属正常、不阻塞。
  6. 是否破坏缓存:不涉及(纯文档改动)。
  7. 是否改了设计方向:不涉及。
  8. 是否引入新配置:不涉及。(唯一一处内容调整:段内交叉引用 CONFIGURATION.md 改为 configuration.md,因为 master 已将该文件移至 docs/configuration.md 且 docs/ 内统一用小写相对链接,避免死链。)

判定:可合入(merge-ready)。

Re-resolve docs/clients*.md after master added a DSH-Go host paragraph at the same anchor: keep master's DSH-Go paragraph, re-apply the #5148 fetch-accessor coexistence note right after it (EN + ZH). Net delta vs current master = only the two #5148 paragraphs.
@ranxianglei

Copy link
Copy Markdown
Owner Author

🤖 Powered by ework · qwen3.8-27b

[bot] 🏷
(ai: vllm-qwen/qwen3.8-27b)

This forward was NOT stale — the conflict genuinely re-appeared, and I re-resolved it.

Why it came back: six days of master activity included PR #6592 (dsh-go docs), which added a “DSH Go host (third-party Android app)” paragraph at the exact same anchor my #5148 note sits at (right after the dsh section’s dependencies. / 不要本地重新推算。, before ## Kimi Code). Two different paragraphs inserted at one spot → a fresh modify-vs-insert conflict against current master.

What I did: merged current master (a259e0b8fa) into the branch, kept master’s new DSH-Go paragraph, and re-applied the #5148 fetch-accessor coexistence note (EN + ZH) immediately after it. Pushed efd7a2dac6 as a clean fast-forward (no force-push), as the unmarked final head (full CI pass, per the CI-hold policy).

Verified live this run:

  • head efd7a2dac6; mergeable: true, mergeable_state: **blocked** (was dirty) — the conflict is cleared; “blocked” is just the normal pre-merge state (branch protection awaiting reviews/CI), not a conflict.
  • Result vs current master = only the two Bug: billion-context fetch accessor prevents dsh experimental-inspector from activating #5148 doc paragraphs (docs/clients.md +19, docs/clients.zh-CN.md +2); master’s DSH-Go paragraph and the pi “Nudge visibility” note are preserved untouched, no other file changed.
  • CI: the delta is docs-only; the prior head was fully green (30 checks, 0 failed) and the new head’s CI is re-running now.

Net: conflict cleared → PR is merge-ready, pending the standard branch-protection gates (reviews + CI). Merging stays human-only.


中文八问:

  1. 是什么问题:该 docs PR 与 base 分支重新产生合并冲突(merge_state 由 behind 变回 dirty)。
  2. 什么原因导致:过去 6 天 master 上的另一个 PR([PR] docs(clients): dsh-go host | docs(clients): dsh-go 宿主 #6592 dsh-go 文档)在 dsh 段落的同一锚点(dependencies. 之后、## Kimi Code 之前)新增了“DSH Go 宿主”段落,与本 PR 的 Bug: billion-context fetch accessor prevents dsh experimental-inspector from activating #5148 共存段落落在同一位置——双方在同一处插入不同内容,于是对当前 master 再次冲突。
  3. 影响范围与严重程度:仅阻塞该 docs PR 的合并,无任何运行时/功能影响。
  4. 怎么修复的:把当前 master(a259e0b8fa)合入分支,保留 master 新增的 DSH-Go 段落,再把本 PR 的 Bug: billion-context fetch accessor prevents dsh experimental-inspector from activating #5148 共存段落(EN+ZH)重新插到它之后、Kimi Code 小节之前;干净快进推送 efd7a2dac6(未强推)。
  5. 是否有回归问题:无。结果相对当前 master 仅新增两段 Bug: billion-context fetch accessor prevents dsh experimental-inspector from activating #5148 文档(docs/clients.md +19 / docs/clients.zh-CN.md +2);master 的 DSH-Go 段落与 pi 段 nudge 说明均原样保留,未触碰任何其他文件。
  6. 是否破坏缓存:不涉及(纯文档改动)。
  7. 是否改了设计方向:不涉及。
  8. 是否引入新配置:不涉及。

判定:冲突已清除(mergeable_state dirty→blocked,mergeable: true),PR 现为 merge-ready,待常规分支保护门(评审 / CI)通过后由 owner 合并。

@ranxianglei ranxianglei left a comment

Copy link
Copy Markdown
Owner Author

Choose a reason for hiding this comment

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

(ai: vllm-qwen/qwen3.8-27b)

Verdict: APPROVE (posted as a comment — GitHub does not allow the owner to formally approve their own PR, so this is a COMMENT-event review carrying the full approval justification).

docs-only, factually accurate, en/zh in sync, CI green, nothing to fix.

I verified the true delta against current master is exactly two files — docs/clients.md (+19) and docs/clients.zh-CN.md (+2); 21 insertions, 0 deletions, no other file touched. Every factual claim in the note checks out against the source:

  • The #1158 self-heal / guarded accessor — src/agent/fetch-chain/index.ts: shouldReclaimFetchPatch() (line 23) reads BILI_RECLAIM_FETCH_PATCH; the accessor descriptor is defined on globalThis.fetch and its setter re-chains any third-party fetch install as downstream, so routing never silently leaves the proxy. The note's description is accurate.
  • Why the accessor stays on by default — with BILI_RECLAIM_FETCH_PATCH=0 the classic direct install applies and a third-party re-arm wins, which is exactly the #1158 silent un-routing the note warns about (the one-time warning at index.ts:258 says so verbatim). Accurate.
  • The cross-reference to #3528 — docs/configuration.md (the BILI_RECLAIM_FETCH_PATCH entry) already carries a "Coexistence note (#3528)" naming this same @deepseek-ai/dsh-experimental-inspector incompatibility. The two docs now point at each other correctly; the clients.md note adds the dsh-specific load-order framing and the captureFetch: false workaround that the config doc doesn't have, so they're complementary, not redundant.
  • Error string, workarounds, load order — the quoted error inspector: globalThis.fetch is an accessor and cannot be observed safely, the three workarounds ordered least→most impact (captureFetch: false / disable Developer Tools / BILI_RECLAIM_FETCH_PATCH=0), and "the profile loads bili before the inspector" all match the #5148 report and the reporter's verified workarounds.

en/zh parity holds (the ZH note is a faithful translation), the note sits in the dsh section immediately before ## Kimi Code, and no runtime code / config schema / wire / persistence is touched. No new config or env is introduced (BILI_RECLAIM_FETCH_PATCH pre-exists and is already documented). CI is green (30 checks, 0 failed, 0 pending). Scoping is correct: it documents the in-scope part of #5148 and leaves the root-cause fix (inspector side, cross-repo) tracked separately pending owner sign-off, so the issue is correctly not auto-closed.


中文八问:

  1. 是什么问题:dsh Developer Tools 启用的 @deepseek-ai/dsh-experimental-inspector 与 bili dsh 原生插件装的受守卫 globalThis.fetch 访问器冲突,inspector 那条 entry 激活失败(dsh web 其余照常)。本 PR 纯文档:在 clients.md 的 dsh 段补一条共存说明。
  2. 什么原因导致的:加载顺序 + 机制冲突——profile 先载 bili(装访问器,#1158 自愈)、后载 inspector(要求 fetch 是普通数据属性,读到访问器就整条抛错)。非 bili 缺陷;关访问器会重新引入 #1158 静默失路由,故默认保持开启。
  3. 影响范围与严重程度:仅文档,零运行时影响、零回归风险。触发条件=同时启用 dsh Developer Tools + bili 原生插件。属体验瑕疵(inspector 一条 entry 不激活,有现成绕行),不影响压缩主链路。
  4. 怎么修复的:文档层面——dsh 段补共存说明:成因(加载顺序+机制冲突,非 bili 之过)、为何访问器默认开(关了回归 #1158)、三条绕行按影响从小到大(① inspector 的 captureFetch: false ② 关 Developer Tools ③ BILI_RECLAIM_FETCH_PATCH=0)、干净修法在 inspector 侧(只降级 fetch-capture 而非整条抛错);en/zh 双语同步,交叉引用 configuration.md 既有 #3528 提示。
  5. 是否有回归问题:无。纯文档,不触碰任何运行时路径;高危模式(语义漂移 / fail-fast 变吞错 / 阈值·时序·输出格式变化)逐项过,均不涉及。特别确认 BILI_RECLAIM_FETCH_PATCH 默认值(on)未动——没有为迁就 inspector 去关默认访问器。
  6. 是否破坏缓存:不涉及。
  7. 是否改了设计方向:不涉及。
  8. 是否引入了新的配置:无新配置 / 新 env。BILI_RECLAIM_FETCH_PATCH 已存在且已在 configuration.md 文档化,本条仅交叉引用。

判定:可合(APPROVE)。 合并操作留给人工。

@ranxianglei

Copy link
Copy Markdown
Owner Author

🤖 Powered by ework · qwen3.8-27b

(ai: vllm-qwen/qwen3.8-27b) review-worker-r1 完成:PR #5151 审毕 → APPROVE(docs-only,真 diff 仅 docs/clients.md +19 / docs/clients.zh-CN.md +2;事实逐条对源码核过——fetch-chain 访问器 #1158、BILI_RECLAIM_FETCH_PATCH 默认值、#3528 交叉引用均属实;en/zh 同步;CI 30 绿 0 红;无小修)。GitHub 侧已发评审(owner 无法正式 approve 自己的 PR,改以 COMMENT 评审承载完整判据 + 中文八问,review id 5946644230)。

This branch has not been deployed

No deployments
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant