Skip to content

[PR] fix: re-anchor fold coverage mutated by host tool-result pruners via byte-verified sentinel shape (#5395) - #5401

Open
ranxianglei wants to merge 1 commit into
masterfrom
2026-11-01_pruner-shape-reanchor
Open

ranxianglei wants to merge 1 commit into
masterfrom
2026-11-01_pruner-shape-reanchor

Conversation

@ranxianglei

Copy link
Copy Markdown
Owner

查重

与 #3849(永久 unmatched 残差 riding wire)同症状家族但机制不同:#3849 的残差来自 wire/dialect 切换后所有身份通道同时失明;本 issue 的残差来自宿主自己的 tool-result pruner 逐字改写了 bili 已折叠消息的正文(head+MARKER+tail 拼接),内容哈希/simhash/wordbag 全部失配,且消息仍在、或然清扫也不触发。两单已互链;#3849 的 absent-retire 兜底走 PR #5398,本 PR 修的是本 issue 的根因(把这类变异变成可认领的)。无其他重复在途 PR。

是什么问题

v0.1.210 plugin lane(dsh desktop 自带 tool-result-pruner 压缩插件)会把 >8KB 的工具结果原地改成 前 4KB + "\n\n[... tool result middle pruned ...]\n\n" + 后 1KB。bili 的消息身份是内容哈希——正文一变,已折叠块覆盖的 id 就永远对不上任何锚点,每个回合都报一次 resent history drifted: N covered id(s) missing with no anchor match,永不收敛(报告者实测 614 ids × 156 次/天)。

修法(A+B,均为 host 侧,不碰 kernel/)

A — 字节验证的 pruner-shape 重锚通道(src/fold-reconcile.ts):reconcile 规划器新增最后一个认领通道。仅当线上出现带已知 sentinel marker 的重发文本时才激活(稳态零存储读取,一次 indexOf/消息);此时只拉取「缺失 covered id ∩ 候选匹配(role/contentType 一致且变短)」的原文(上限 64),逐字节验证 newText = original[0..head] + MARKER + original[-tail..](非空中间删除 + 重叠守卫)才认领。这是最强证据等级:重发字节字面分解为已知拼接形状且头尾与存储原文逐字相等,误绑概率≈0。故意不做 toolCallId 相等门——宿主跨投影重铸 id(#2396 类)正是永久漂移成立的条件之一。marker 表内置常量(当前仅 dsh pruner 一个),不开配置面(配置面 owner-gated;新宿主接入走 PR)。

B — 冲突台账可见性(src/conflict-watch.ts):新增 ConflictKind host-content-mutation,byPrune>0 时记一条(字节级实证的第二压缩器证据),面板冲突 banner/advice 首次能显示「双压缩器」状态而非只有日志。自限:repair 模式重锚后下一 pass byPrune 归零。

行为变更披露(旧→新)

项 旧 新 原因
host-pruner 变异后的折叠覆盖 每回合永久 drift warn + 残差 token 反复上 wire 一 pass 内重锚到新 id,收敛为干净 resend 本 issue 根因修复
FoldReconcileResult 无 byPrune 字段 新增 byPrune: number(additive) 可观测性
ConflictKind 7 种 新增 host-content-mutation(additive) 双压缩器可见性(报告者问 B)
稳态开销 — 无 marker 时零额外存储读取(仅子串扫描) 设计约束

无配置面变更、无 kernel/ 变更、无 wire 形状变更、无默认值变更。

两种压缩模式

通道在 planReconciliation 核心内运行,plugin lane 与 proxy lane 共用同一 reconcile 路径,行为一致;测试含 proxy(pluginAgent undefined)等价用例。

验证

  • npm run typecheck clean(rebased head)
  • 定向套件 186/186 本地通过:tests/issue5395-pruner-shape.test.ts(14 个:回归复现=重铸 id+pruned 字节→byPrune=1 收敛;负向门×5:外部原文/无 marker/增长/role 失配/存储不可用;k-th 一对二;warn 模式计数不改块;proxy 等价)、tests/fold-reconcile.test.ts、fold-drift、fold-round2-shape、fold-anchor-doctrine、fold-reconcile-positional、issue1206-thirdparty-scan、issue5142-expected-rebase(master 同期改动回归)、issue4732-fold-decomposition-dedup
  • 全量回归以远程 CI 为准
  • MESSAGE-IDENTITY.md 增补「verified mutation class」一节(该文档原先声明 fuzzy join 被拒、此类验证通道是 open follow-up)

Review 提示

身份认领逻辑按 §7.4 属人工 review 层(message-ref/id logic)。与 master #5142(expected-rebase accounting,同文件热区)已 rebase 合并:两者正交——#5142 让「合法退役的 id」退出 missing,本 PR 让「被宿主改写的 id」重新进入认领,参数各占一位(expectedAbsent, storedById)。

Closes #5395

@ranxianglei ranxianglei changed the title fix: re-anchor fold coverage mutated by host tool-result pruners via byte-verified sentinel shape (#5395) [PR] fix: re-anchor fold coverage mutated by host tool-result pruners via byte-verified sentinel shape (#5395) Nov 1, 2026
@github-actions

github-actions Bot commented Nov 1, 2026 •

Copy link
Copy Markdown
Contributor

📦 Built Package Artifact

Branch: 2026-11-01_pruner-shape-reanchor (288f2e2)

Latest build — install this exact version

npm install -g billion-context@0.0.8937-pr.5401.533fe60

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
288f2e2 billion-context@0.0.8937-pr.5401.533fe60 build
fa6bcd0 billion-context@0.0.7641-pr.5401.7b5fcb3 build

Download artifact

  1. Download the artifact from the Actions run
  2. Extract the tarball and install:
tar xzf billion-context-pr5401.tgz
npm install -g package

Install instructions are refreshed on each push.

@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 PR 是 #5395 的实现(Closes #5395)。相关条目已核对:

编号对照说明:镜像 issue #5403 对应的是 GitHub PR #5401(上游不存在 refs/pull/5403/head;#5401 head fa6bcd08b 标题与本条完全一致,parent = 当前 master 7578e8184,单提交、base 最新)。

审核启动中:身份认领逻辑按 §7.4 属人工 review 层。已独立通读全量 diff(6 文件,+518/−13:src/fold-reconcile.ts / src/conflict-watch.ts / src/fold-decomposition-lookup.ts / MESSAGE-IDENTITY.md / 2 个测试文件),回归清点与 CI 状态核查进行中,八问结论随后补上。

@ranxianglei

Copy link
Copy Markdown
Owner Author

🤖 Powered by ework · qwen3.8-27b

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

PR #5401(镜像 #5403)独立审核结论

标签: bug dsh scope/kernel p2(缺陷类;dsh 桌面车道实测;src 侧折叠判定面属 scope/kernel;能用但每回合持续残差上 wire+drift warn、永不收敛 → p2,非数据丢失/缓存泄露)。

八问

  1. 是什么问题:dsh desktop 自带的 tool-result pruner 把 >8KB 工具结果原地改写为「前 4KB + sentinel marker + 后 1KB」。bili 消息身份是内容哈希,正文一变,已折叠块覆盖的 id 永远对不上任何锚点——每回合报一次 resent history drifted: N covered id(s) missing,永不收敛(报告者实测 614 ids × 156 次/天、~571K tok/天残差反复上 wire),且冲突台账恒 0,「双压缩器」状态完全不可见。
  2. 什么原因导致的:不是某条通道的 bug,而是身份 ladder 对这类变异的结构性盲区——「原文仍在、正文被 head+MARKER+tail 拼接」的形状让所有精确通道(内容哈希/顺序档/simhash/wordbag/toolCallId)同时失明;消息仍在,absent-retire([Bug] Mid-session wire/dialect switch leaves a permanent unmatched-id residual that rides the wire unfolded and taxes the prefix cache every turn (v0.1.201) #3849/[PR] fix: retire persistently-unmatched fold coverage after 3 evidence-less passes (#3849) — relay of stranded branch #5398)也不触发。ladder 此前没有为这类变异设任何认领通道。
  3. 影响范围与严重程度:带宿主侧 pruner 类插件的 plugin 车道(当前仅 dsh desktop 实测);严重度 = 能用但有持续成本与噪音(每回合残差 token 反复上 wire + 一条 drift warn,永不收敛),非数据丢失、非缓存只增不减 → p2。
  4. 怎么修复的:A+B 双管,均在 host 侧,kernel/ 未碰。A = reconcile 规划器新增最后一个认领通道:仅当线上出现含已知 sentinel marker 的重发文本才激活(稳态零存储读取,仅子串扫描),对「缺失 covered id ∩ role/contentType 一致且变短的候选」拉统一存储原文(上限 64),逐字节验证 newText = original[0..head] + MARKER + original[-tail..](非空中间删除 + 重叠守卫)才认领——最强证据等级(重发字节字面分解为已知拼接形状、头尾与存储原文逐字相等,误绑概率≈0)。故意不设 toolCallId 相等门(宿主跨投影重铸 id [Bug][Prime/Responses] provider switch rewrites composite tool IDs and reintroduces compressed history #2396 正是永久漂移成立的条件之一)。marker 表内置常量,不开配置面(owner-gated 纪律)。B = ConflictKind 新增 host-content-mutation,byPrune>0 记一条,面板冲突 banner/advice 首次能显示双压缩器状态。
  5. 是否有回归问题:既有行为逐项清点——planReconciliation 全仓唯一调用者是 reconcileFoldCoverage,新参可选、其他路径惰性;FoldReconcileResult 全部 6 个返回点已更新(grep 验证)、测试 deepEqual pin 同步更新;expectedAbsent 交互干净([Performance][dsh] Mega-sessions pay an ever-growing O(full-history) fold-reconcile tax every turn: expected host churn (one-shot carrier retirement, image/offload retroactive rewrites) misclassified as drift | 大会话每轮付不断增长的 O(全史) fold-reconcile 税:合法宿主变更(o… #5142 合法退役 id 不会被新通道认领,早退先于新通道);无 marker 稳态零额外存储读取、行为零变化;高危模式逐项过(fail-fast→吞错 / 阈值漂移 / 时序 / 输出格式 / wire 形状)——均未命中。两个不阻塞发现:①理论边缘——若未来某 marker 变体为空 tail(marker 位于文本末尾),验证退化为前缀-only,k-th 纪律把伤害限定为一候选一 id(最坏=错 id 被重锚、正主骑线 unfolded,是成本不是损坏);当前内置表仅 dsh 默认形状(tail=1024),现有常量下不可达;②recordConflict 无去重,持续 warn 模式(非默认)下台账每轮记一条(环上限 20 内自限),repair 默认模式一轮收敛(测试已钉住)。
  6. 是否破坏缓存:不破坏。重锚只是把块覆盖指到客户端新 id,重发字节保持客户端逐字(cache-stable),未触碰前缀缓存锚逻辑。
  7. 是否改了设计方向:没有。ladder 尾部加一个通道;判定/执行共用投影([Bug] recommend/apply 对同一区间的 turn-integrity 判定永久不一致:acp_status 反复推荐的区间被 apply 侧整段拒绝(0.1.196,Range would split N turn(s) at the protected-zone boundary) #3236)、msgid 钉身份(Plugin lanes: pin identity to client msgids, render outbound from the server-owned history | 插件车道服务端正史:消息身份钉在客户端 id 上,出站前缀由正史渲染 #4491/[PR] fix(pi-msgids): model-switch rewrite no longer drops client message ids (pairing proof preserved) | fix(pi-msgids): 切模型改写不再丢客户端消息 id(配对证明保留) #4834)、快照渲染车道均不受影响;MESSAGE-IDENTITY.md 增补「verified mutation class」一节记录该教义边界。
  8. 是否引入新的配置:没有。无 env/config/默认值;marker 表为内置常量(新宿主接入走 PR,配置面 owner-gated 纪律遵守)。

验证证据(独立产出,未采信 issue 侧结论):

与在途单关系:#5398(#3849 的 absent-retire 有界出口)正交互补——本 PR 让「被宿主改写的 id」重新可认领,#5398 让「真消失的 id」有界退出,合并顺序无依赖。

判定:ready-to-merge。身份认领逻辑按 §7.4 属人工 review 层(message-ref/id logic),以上为我的独立审核结论,合并由 owner 决定。

@ranxianglei ranxianglei added bug Something isn't working dsh p2 scope/kernel labels Nov 1, 2026

@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: ready to merge (approve-equivalent). Posted as a comment-review — GitHub refuses to record a formal approval on a self-authored PR. Independent end-to-end read of the full diff (6 files, +518/−13). No kernel/ touch, no config-surface change, no wire-shape change. Findings below are my own trace, not a restatement of the issue thread.

Correctness (verified against surrounding code, not just the hunks):

  • verifyPruneShape (src/fold-reconcile.ts) is an exact decomposition: head+marker+tail where head/tail are byte-exact prefix/suffix of the stored original, non-empty deleted middle enforced by original.length >= headLen + tail.length + 1 (which also kills overlapping prefix/suffix false-matches), trying every marker occurrence (the double-prune case). Byte-level identity ⇒ mis-attribution is structurally excluded.
  • Runs LAST in planReconciliation (after every exact pass + the #3641 shape pass), so it can't steal a claim a stronger channel made — pinned by the "stable toolCallId ⇒ byTool=1, byPrune=0" control test.
  • k-th discipline holds: backbone order; one oldId→≤1 candidate (break); one candidate→≤1 oldId (claimedCandidates). The shared-head/tail test pins "one candidate serves at most one old id."
  • #5142 interaction clean: missingSet excludes expectedAbsent ids at construction (!expectedAbsent?.has(id)), so legitimately-retired ids cannot be re-anchored here.
  • Zero steady-state storage reads: originals are pulled only when a marker-bearing resend exists, and only for superset-filtered missing ids (cap 64); no marker on the wire ⇒ index untouched (contract pinned by a test).
  • Reasoned in BOTH compression modes: the pass lives in the shared planner/core with no pluginAgent-dependent branch, so plugin and proxy lanes behave identically.

Regression inventory (existing behaviors the diff touches, each judged):

  • All 5 early returns + the final return of reconcileFoldCoverage carry the new byPrune (additive field complete; no partial result shape).
  • FoldReconcileResult.byPrune and ConflictKind.host-content-mutation are purely additive; the existing deepEqual pin in tests/fold-reconcile.test.ts was updated to match.
  • Drift log lines gain one optional clause (inert when byPrune == 0); the conflicts banner gains a first-match-wins advice tier (mixed ledgers keep their existing tiers verbatim).
  • The re-anchor only repoints block coverage onto the client's own resent id — bytes stay client-verbatim ⇒ prefix-cache stable; the one-time unfold→summary transition is the intended fix and is disclosed old→new in the PR body. No fail-fast→swallow, no threshold/timing/output-format drift, no cache-invalidation-strategy change.

Tests: genuine regression repro (re-minted toolCallId + pruned bytes ⇒ byPrune=1, converges in one pass) plus #3617 both-directions gates (foreign-original / no-marker / growth / role-mismatch / storage-absent each assert byPrune=0 + honest unmatched), cross-pass convergence (the bug was NON-convergence; the second pass is a clean resend with no ledger spam), warn-mode parity, k-th discipline, and the zero-read contract. Assertions check specific claim mappings / counts / state transitions — not tautologies.

CI: green on head fa6bcd08b (30 checks, 0 failed, 0 pending).

One non-blocking note (pre-existing, NOT introduced by this PR, benign): the #3641 shape pass claims ids after the plan.unmatched freeze but does not drop them from plan.unmatched (this new pass does, at src/fold-reconcile.ts:1657). It is harmless because any shape claim sets byShape > 0, which both forces shapeExplained and zeroes totalDrift, so shape-claimed ids never reach the destroy verdict. Left as-is deliberately (changing it would alter already-tested #3641 behavior); flagged for completeness only.

Merge stays human per house rules.

@ranxianglei

Copy link
Copy Markdown
Owner Author

🤖 Powered by ework · qwen3.8-27b

(ai: vllm-qwen/qwen3.8-27b) Review complete — verdict ready to merge. Independent end-to-end read of the full diff (6 files, +518/−13): no kernel/ touch, no config-surface change, no wire-shape change; the byte-verified pruner-shape re-anchor (verifyPruneShape exact head+marker+tail decomposition, runs LAST so it can't steal stronger channels, k-th discipline, clean #5142 expectedAbsent interaction, zero steady-state storage reads) verified correct against surrounding code, reasoned in both compression modes; tests carry real teeth (regression repro + #3617 both-directions gates + cross-pass convergence + warn/degradation/zero-read contracts). One pre-existing benign note logged (the #3641 shape pass leaves its claims in plan.unmatched; harmless because byShape>0 masks the verdict) — non-blocking. CI green on head fa6bcd08b (30 checks, 0 failed). Posted as a comment-review since GitHub blocks approving a self-authored PR: #5401 (review)

…byte-verified sentinel shape (#5395)

Host compression plugins (dsh desktop's tool-result-pruner) splice a
versioned sentinel between preserved head and tail of long folded tool
results. The middle deletion breaks every identity channel at once (norm/
twin equality, wordbag/simhash distance, positional fence); when the host
also re-mints tool-call ids across projections (#2396 class) the covered id
goes PERMANENTLY unmatched — re-reported every turn, riding unfolded and
feeding #2438/#5223 escalation (reported: 614 ids x 156 turns/day).

A: pruner-shape pass — LAST channel in planReconciliation. Claims only on
positive evidence: resent bytes decompose as head + sentinel + tail with
head/tail byte-matching the STORED original (unified index), non-empty
deleted middle (overlap guard). No toolCallId equality gate (re-minting is
exactly the permanent-drift condition). Steady state pays zero storage
reads (marker substring scan first; rows pulled only when a sentinel is on
the wire). Store unavailable -> honest unmatched.

B: each verified claim lands one 'host-content-mutation' conflict-ledger
event with its own advice tier — the second compressor becomes visible on
banner/stats instead of staying log-only. Self-limiting: repair mode
re-covers the pruned ids, next pass sees them present.

Behavior change (intended, the bug fix): pruned covered ids were previously
permanently unmatched (drift warn every turn forever); they now re-anchor
in ONE pass and converge to clean resend. Additive FoldReconcileResult.byPrune
field; additive ConflictKind member; no config surface, no kernel/, no wire
shape change. Runs identically in plugin and proxy lanes.
@ranxianglei

Copy link
Copy Markdown
Owner Author

🤖 Powered by ework · qwen3.8-27b

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

Both blockers cleared — PR is conflict-free and fully green at head 288f2e2a2c.

1. Merge conflict (previous wake, resolved this session). Rebased fa6bcd08b onto master 4dea18eabf. Exactly one textual conflict: MESSAGE-IDENTITY.md — master turned that file into a redirect stub when docs centralized under docs/ (#5920), while the PR appended its new #5395 section to the old root path. Resolution followed the doc-move rule: stub kept as-is, and the PR's "verified mutation class" section moved verbatim into docs/message-identity.md at the same relative position (after the #3641 compatibility paragraph, before the #3756 section). src/conflict-watch.ts and src/fold-reconcile.ts auto-merged; I verified both sides survived intact in the merged tree — master's #5715 hardDirectMajority logic and the PR's byPrune pruner-shape pass coexist, with the pruner pass still running LAST (before the #3641 residual census), so no semantic reordering. Local verification on the rebased head: typecheck clean; targeted suites green (issue5395-pruner-shape + fold-reconcile: 86/86; conflict-lifecycle/conflict-remediation/fold-drift/web-conflict-banner: 52/52). Pushed with --force-with-lease pinned to the exact prior sha (explicitly sanctioned for this conflict rebase); remote head was unchanged at push time.

2. Red CI on the rebased head (this wake). Live-verified first: 1 of 30 checks red — test (windows-latest, 22, 2), single failing test abort x proxy-openai-sse: client abort mid-stream destroys the upstream request (tests/wire-exit-matrix.test.ts:273, closed=0, duration 21.7s; sibling cells in the same run passed at 140/71/432ms). Triage: this exact cell + shard is a documented recurring load-starvation flake — the test file header records the full budget history (3s→6s→10s→win32 20s→lag-extension cap) with every recorded instance being windows-latest node22 shard 2, closed=0, siblings green, last recorded 2026-11-04 (run 41776902208); the mandated teardown audit found no real hang (relay arms close → abort() synchronously, no commit since d244ffff touched destroy/abort wiring). The PR diff touches none of that path (only fold-reconcile / conflict-watch / fold-decomposition-lookup / docs / tests). Action: triggered rerun-failed-jobs — attempt 2 completed success. No code change made or needed; branch left at 288f2e2a2c. One-line conclusion posted to #4599 per CI-coordination policy.

Live state now (verified this run): all 48 check runs on 288f2e2a2c = 36 success + 12 skipped, zero failures/pending; mergeable: true (state "behind" only because master advanced past the rebase base — no conflict).


结构化八问(中文)

  1. 是什么问题:PR 与 master 冲突(文档文件被集中迁移成重定向桩)+ rebase 后 Windows node22 shard 2 一个 teardown 测试单红。
  2. 什么原因导致的:冲突根因是 master 侧 [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 把根目录文档移到 docs/ 并留桩,PR 仍往旧路径追加章节;CI 红是 wire-exit-matrix abort×openai-sse cell 在加载型 Windows runner 上的已知事件循环饥饿 flake(文件头注释有完整阶梯史),与 PR 改动无关。
  3. 影响范围与严重程度:冲突不解决则 PR 无法合并(阻塞级);CI 红为单 cell flake,同一代码 rerun 即绿,非回归、无用户面影响。
  4. 怎么修复的:冲突按文档迁移规则解——保留桩、把 PR 新增章节逐字落到 docs/message-identity.md 对应位置;自动合并的两个源文件逐一核验两侧改动([Bug][DSH] Partial interior loss is classified as total drift despite surviving direct-ID majority, latching compression off #5715 与 byPrune 通道)都在且顺序未变;CI 红走 rerun-failed-jobs 验证为 flake,不动代码。
  5. 是否有回归问题:无。rebase 后 typecheck 干净、定向套件 138/138 绿;pruner-shape 通道仍在 planReconciliation 末位运行,[Bug][DSH] Partial interior loss is classified as total drift despite surviving direct-ID majority, latching compression off #5715 多数派判决逻辑未被扰动;未改任何 wire/config/持久化行为。
  6. 是否破坏缓存:不涉及——未触碰缓存、重试、游标或去重键相关代码。
  7. 是否改了设计方向:不涉及——纯冲突消解 + flake 处置,无架构/数据流变更。
  8. 是否引入了新的配置:不涉及——无新 env/config/默认值。

判定:冲突已解、CI 全绿、mergeable=true,PR 处于可合并状态,等待人工合并。

@ranxianglei

Copy link
Copy Markdown
Owner Author

🤖 Powered by ework · qwen3.8-27b

[bot] 🏷 (ai: vllm-qwen/qwen3.8-27b)

Confirmation close-out (full detailed summary already posted above, comment id 70437):

Verdict: both blockers cleared, PR is conflict-free and fully green — ready for human merge.

@ranxianglei

Copy link
Copy Markdown
Owner Author

🤖 Powered by ework · qwen3.8-27b

[bot] 🏷 (ai: vllm-qwen/qwen3.8-27b)

Re-verified live this run — no action needed: the reported test (windows-latest, 22, 2) failure is the original attempt 1 of run 42817946339, which was already handled in the previous wake (documented load-starvation flake in the wire-exit-matrix abort×proxy-openai-sse cell; rerun-failed-jobs attempt 2 completed success, verified directly against the Actions API then). Mirror check cache fetched just now (15:22Z) on head 288f2e2a2c: 30 checks, 0 failed, 0 pending — all green. No code change made or needed; branch stays at 288f2e2a2c, ready for human merge.

@ranxianglei

Copy link
Copy Markdown
Owner Author

🤖 Powered by ework · qwen3.8-27b

[bot] 🏷 (ai: vllm-qwen/qwen3.8-27b)

七问(本轮 CI 处置摘要;前两轮完整记录见上方 comment id 70437)

  1. 本轮报告了什么:head 288f2e2a2c 上 test (windows-latest, 22, 2) 红(run 42817946339)。
  2. 活体复核结果:该红项是该 run 的 attempt 1 原始失败;上一轮已判定为已知 flake(wire-exit-matrix abort×proxy-openai-sse cell 在 Windows node22 shard 2 的事件循环饥饿,closed=0 @21.7s、同跑兄弟 cell 全绿、PR diff 与该路径零交集)并触发 rerun-failed-jobs,attempt 2 已直接对 Actions API 验证为 success。本轮再读镜像缓存(15:22Z 抓取):30 checks, 0 failed, 0 pending,全绿。
  3. 是否改了代码:否——无需任何机械修复,未动分支。
  4. 分支状态:仍为 288f2e2a2c,本轮无 push(无并发冲突风险)。
  5. 回归风险:无(零改动)。
  6. 是否破坏缓存:不涉及。
  7. 结论:上游阻塞项已全部清除且经活体验证,PR 处于可合并状态,等待人工合并。

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

bug Something isn't working dsh p2 scope/kernel

Projects

None yet

Development

Successfully merging this pull request may close these issues.

[Bug] compressible-range chars counts only text, so signature-only thinking runs size as 0 and the ?? tokens*4 fallback never fires

1 participant