Skip to content

[PR] fix: log compress applications onto reused ref numbers (#5199) - #5212

Open
ranxianglei wants to merge 9 commits into
masterfrom
2026-10-30_restamp-dual-identity-log
Open

ranxianglei wants to merge 9 commits into
masterfrom
2026-10-30_restamp-dual-identity-log

Conversation

@ranxianglei

Copy link
Copy Markdown
Owner

Related to #5199; direction decision pending at #5204 (this PR is the no-decision-dependent part of options A/C).

What and why

重启后冷加载的 ledger 重排可能把已发放的 ref 重新发给另一条消息(空置号在同一次扫描里作为别人的 seq+1 目标被占用)。守卫 live-first:引用旧号的压缩调用会被静默地应用到新持有者并返回成功——#5199 用隔离 fixture 复现了这个最坏形态。

live-first 仍是接受的裁决(拒绝会死循环,#4797;翻译成旧持有者会系统性误判合法的 post-restamp 引用)。本 PR 不改任何接受行为,只把原本静默的应用变为可见:

第二个 commit(标注的顺手 docs 修):修正两处与代码事实不符的注释——reference/ref-keyed-surfaces.md(contract 段把已 revert 的 prune 写成 enforced;「已消除其全部用户可见危害」断言被 #5199 证伪)与 tests/issue4797-superseded-deadlock.test.ts 头注释(描述的是 post-prune 的 superseded 形状,master 实际产出 {m3→b, m6→e, m7→f},其中 m6 即双身份案例)。

Behavior changes

无。接受/拒绝/翻译行为逐字节不变;仅新增一条服务端日志行(且只在双身份真实发生时出现)。

Verification

  • npm run typecheck clean
  • 定向测试:tests/issue5199-dual-identity-log.test.ts(新,4 例:双身份应用折叠当前持有者+告警触发 / 负对照无告警 / 去重与 fast-path 契约 / 翻译路径不受影响)+ tests/issue4797-superseded-deadlock.test.ts(既有 4 例仍绿)
  • 全量套件交给 CI

Boundaries

@github-actions

github-actions Bot commented Oct 30, 2026 •

Copy link
Copy Markdown
Contributor

📦 Built Package Artifact

Branch: 2026-10-30_restamp-dual-identity-log (3de730d)

Latest build — install this exact version

npm install -g billion-context@0.0.9206-pr.5212.f7a78fd

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
3de730d billion-context@0.0.9206-pr.5212.f7a78fd build
4f279f7 billion-context@0.0.7187-pr.5212.f71628e build
fde2f0d billion-context@0.0.7072-pr.5212.a116f7d build
2fb797c billion-context@0.0.7067-pr.5212.260fb07 build
4b7695e billion-context@0.0.7048-pr.5212.676db59 build
7f5ebbf billion-context@0.0.6999-pr.5212.00753e6 build
ddf31aa billion-context@0.0.6989-pr.5212.da169c9 build

Download artifact

  1. Download the artifact from the Actions run
  2. Extract the tarball and install:
tar xzf billion-context-pr5212.tgz
npm install -g package

Install instructions are refreshed on each push.

ci-check and others added 4 commits October 30, 2026 18:00
After a restart, the cold-load ledger restamp can hand a pre-restart ref
to another message (a vacated number re-issued as someone else's seq+1
target in the same pass). The guard resolves endpoints live-first, so a
replayed pre-restart compress call citing the old number applies to the
new owner and returns success — silently folding the wrong message.

Live-first stays the accepted resolution (rejecting dead-loops like
#4797; translating to the old owner would mistranslate legitimate
post-restamp cites). This change makes the application visible instead
of silent: supersededRefRejection now reports every cited ref that is
BOTH live and still another message's recorded pre-restamp label
(dualIdentity), and applyRanges logs a prominent warning naming the
refs with salted owner ids only (#1454 pattern). Acceptance behavior is
unchanged — zero semantic change; direction decision pending #5204.
Labeled follow-up fixes for comments that contradicted code facts after
#4807's kernel liveness prune was reverted:

- reference/ref-keyed-surfaces.md: the contract section listed the
  reverted prune as enforced, and the 暂缓 section claimed host
  live-first '已消除其全部用户可见危害' — disproven by #5199 (isolated
  fixture reproduced the silent wrong-target application). Both now say
  what is actually true and point at #5199/#5204.
- tests/issue4797-superseded-deadlock.test.ts: header comment described
  the post-prune superseded shape ({m3→b, m7→f}); master produces
  {m3→b, m6→e, m7→f} where m6 is the dual-identity case now pinned in
  tests/issue5199-dual-identity-log.test.ts.
…HARD_MAX_BYTES

#4788 D2 added the per-session resident hard cap (config field
sessions.hardMaxBytes / env BILI_SESSION_RESIDENT_HARD_MAX_BYTES) but never
registered the surface: no YAML seed entry, no LIFECYCLE.md row, and a
hand-edited EN-only CONFIGURATION.md row that diverged from the generator.
That tripped config-surface-guard, docs-coverage, lifecycle-ledger-guard and
docs-sync on master.

Add the seed entry (restoring the author's original EN wording + the #4792
decision ref + honest dynamic default), the matching ZH rows, and the
LIFECYCLE.md §2 ledger entry, then regenerate the config docs so the artifact
matches the seed again.
…k test

The #4666 suite's terminal-stamp asserts were moved to v9 when the schema
advanced (commit 'expect the v9 stamp in the #4666 migration suite'), but the
concurrent-boots lock test's final assert was missed and still expected v8.
With MYSQL_SCHEMA_VERSION now 9 (#4796 zero-ref index — index-only, no row
change), the migrator legitimately stamps 9 after the lock frees, so
'migration completed after the lock freed' asserted 9!=='8'. Bump to 9 to
match the three sibling asserts; the lock-serialization discriminators
(mid-flight stamp 7, rowCount 8, dedup collapse) are untouched.
@ranxianglei
ranxianglei force-pushed the 2026-10-30_restamp-dual-identity-log branch from 4b7695e to 2fb797c Compare October 30, 2026 10:13
@ranxianglei ranxianglei added bug Something isn't working p1 scope/kernel labels Oct 30, 2026
@ranxianglei

Copy link
Copy Markdown
Owner Author

🤖 Powered by ework · qwen3.8-27b

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

结论:本次唤醒是过期转发——所指 head 7f5ebbf9ba 是悬空 merge commit(不在任何 ref 上),其 run 已被取代。实际处理:已把 PR 分支同步到最新 master + ci-fixes 并推送(4f279f73c),收进哨兵刚落的 #5140 pin 修复;唯一剩余阻塞是 #4788 born-red(master 车道,已在 #4599 跟踪)。

  1. 是什么问题:转发指向 head 7f5ebbf9ba / run 41015459247,但该 commit 是早期 worker 从旧 base(ddf31aa)本地合出的悬空 merge(父 ddf31aa + ci-fixes@9c186cc56,不在任何远端 ref 上),不是当前 PR head——过期事件。检查时真正的当前 head fde2f0d31 的 run 零红(12 success + 4 skipped,其余在跑)。
  2. 什么原因导致的:两层——①转发本身滞后于分支真实状态(镜像检测到的是孤儿 head 的旧 run);②真实的残余红风险仍是继承项:[Bug][fork] Cold restore cannot use an intact unified raw tail for a non-retained parent (#4266 item 4) #5140 cold-tail 409 flake 家族(哨兵已在 ci-fixes 以 67f8e8940 修复)+ [Performance] v0.1.208 long sessions exceed 2 GiB memory and spend 14–22 seconds in local processing #4788-resident-minimal born-red hang(最新 master 44067a5bc 上今日复验仍挂,[PR] fix(#5140): name the failed raw-tail reconstruction gate instead of failing closed silently #5188 未覆盖)。
  3. 影响范围与严重程度:只挡本 PR 的 CI 门禁;不影响产品行为——PR 本体是 visibility-only 日志,零语义变更。
  4. 怎么修复的:PR 代码零改动。把分支同步到 master 44067a5bc(含 [PR] fix(#5140): name the failed raw-tail reconstruction gate instead of failing closed silently #5188 sidecar 恢复修复)+ ci-fixes 67f8e8940(含 [Bug][fork] Cold restore cannot use an intact unified raw tail for a non-retained parent (#4266 item 4) #5140 测试 pin),两次合并均干净,推送 4f279f73c,新 run 已触发。[Performance] v0.1.208 long sessions exceed 2 GiB memory and spend 14–22 seconds in local processing #4788 born-red 已在 CI Control Center: Centralized CI Failure Handling & Status Sync (Comment all CI issues here; do not open new issues) | CI 总控: CI 失败集中处理与状态同步(所有 CI 问题在此评论, 勿开新 issue) #4599 二次上报(带最新复验),等哨兵把修复收敛进 ci-fixes。
  5. 是否有回归问题:无。合并进来的全部是已人审合入的 master 内容([PR] fix(#5140): name the failed raw-tail reconstruction gate instead of failing closed silently #5188)与哨兵收敛的 ci-fixes 内容;定向验证:[Performance] v0.1.208 long sessions exceed 2 GiB memory and spend 14–22 seconds in local processing #4788 测试在同步前后行为一致(两侧同挂 = 继承实锤,与本 PR 无关);PR 自身 diff 维持纯日志(dualIdentity 记录 + 一条 ctx.log)。
  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:不涉及。

为何不开 decision-request issue(attempt 2/2 逃生口):剩余阻塞(#4788 hang)不是设计决策,而是 master 车道的未修测试缺陷——按 owner policy 2026-10-24,一切 CI 问题收敛到 #4599,禁止为 CI 问题开新 issue;该缺陷已带签名+复现+嫌疑提交(1688fc644)两次上报(#4599 评论 6520655388 / 6527216810)。

收工判定:自动修复车道用尽(2/2)。PR 代码无待办,分支已与 master/ci-fixes 完全对齐;哨兵把 #4788 修复收敛进 ci-fixes 后,再做一次同步合并即可全绿。

@ranxianglei

Copy link
Copy Markdown
Owner Author

🤖 Powered by ework · qwen3.8-27b

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

补一条(完整结论见上一条 bot 评论,不重复):本轮已处理完毕——转发所指 head 7f5ebbf9ba 为悬空 commit(过期事件);PR 分支已同步 master + ci-fixes 并推送 4f279f73c(收进 #5140 pin 修复),新 run 已触发;唯一剩余阻塞是 #4788 born-red(master 车道,已在 #4599 两次上报)。自动修复车道 2/2 用尽,PR 代码无待办。

@ranxianglei

Copy link
Copy Markdown
Owner Author

🤖 Powered by ework · qwen3.8-27b

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

Tracking note from the #5207 decision thread (owner ruling 2026-10-31): when #5360 (slot-retirement, kernel 0.0.174) merges, this branch needs a small rebase — the src/stream.ts docblock (~L861-866) still claims the withdrawn kernel prune removes the reused-ref ambiguity. Post-#5360 the correct statement is: restamp retires freed slots within an alignment pass, so new dual-identity numbers are no longer constructed; this PR's log line then covers only the residual cases — pre-fix sessions loaded before the kernel change, and displacement (stamped-over occupant) shapes inside the documented #4807 replay-window tradeoff.

ci-check added 2 commits November 12, 2026 02:36
Resolve base-move conflicts for PR #5199 against current master:

- src/stream.ts: reconcile the residual-hazard comment — kernel #5196 slot-retirement closed the source-level path (pending-decision #5204 superseded), pre-#5196 sessions still carry the ambiguity that dualIdentity now logs.

- tests/issue4797-superseded-deadlock.test.ts: take master's #5196-accurate post-restamp comment (f parks on m8), keep the cross-ref to the issue5199 dual-identity test.

- src/loop/adapter-anthropic.ts, tests/plugin-fork.test.ts: not part of #5199's intent; converge fully to master (inherited #5179/#4599/#5140/#5608 divergence).
…ernel

The PR's original fixture drove its shape through restampRefsToLedger, but
#5196's slot-retirement now parks a colliding fresh stamp on a brand-new
number instead of reusing the vacated one, so byRef.m00006 no longer
resolves to "f" and dualIdentity comes back empty. The host-side
dualIdentity detection + warning log (#5199) still needs coverage for the
sessions whose PERSISTED superseded map predates #5196, so hand-build that
exact end-state (byRef.m00006="f", restampSuperseded={m3->b,m6->e,m7->f})
directly rather than regenerating it through the kernel.
@ranxianglei

Copy link
Copy Markdown
Owner Author

🤖 Powered by ework · qwen3.8-27b

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

Base-conflict resolved and pushed. This run I merged current origin/master into the PR branch, resolved every conflict, fixed one new test that the already-shipped kernel slot-retirement (#5196) had turned red, and fast-forward-pushed (no force). GitHub now reports mergeable: true (was false / dirty).

中文七问:

  1. 是什么问题:PR [PR] fix: log compress applications onto reused ref numbers (#5199) #5212 与 master 基线冲突(mergeable = dirty);并且它新增的「双身份日志」测试在 master 已合入的 [Bug] compress can silently fold another message after restart when a pre-restart ref is reused (0.1.209) #5196 kernel「槽位退休」机制下会转红——本次要一并处理的就是「接冲突 + 修红 CI」这两件事。

  2. 什么原因导致的:分支落后 master 约 1700 个提交。4 个冲突文件里只有 2 个(src/stream.ts、tests/issue4797-superseded-deadlock.test.ts)真正落在 429 governor hold is keyed by origin+credential with no model dimension — one model's quota exhaustion freezes the whole lane (healthy models get local 429/424) #5199 的改动路径上,而且都只是注释块;另外 2 个(src/loop/adapter-anthropic.ts、tests/plugin-fork.test.ts)是分支继承来的 ci-fixes/CI Control Center: Centralized CI Failure Handling & Status Sync (Comment all CI issues here; do not open new issues) | CI 总控: CI 失败集中处理与状态同步(所有 CI 问题在此评论, 勿开新 issue) #4599 接线改动,与 429 governor hold is keyed by origin+credential with no model dimension — one model's quota exhaustion freezes the whole lane (healthy models get local 429/424) #5199 无关。至于 issue5199 的新测试:它用 restampRefsToLedger 现场造双身份夹具,但 [Bug] compress can silently fold another message after restart when a pre-restart ref is reused (0.1.209) #5196 之后该函数会把撞号的新戳停在一个全新号上(m8)、不再复用空出来的 m6,于是 byRef.m00006 不再指向 f、dualIdentity 为空,T1/T3 断言失败。

  3. 怎么处理(修法思路):把当前 master 合并进来后逐文件解冲突——两个与本 PR 无关的文件整文件收敛到 master(纯机械、安全);两个落在 429 governor hold is keyed by origin+credential with no model dimension — one model's quota exhaustion freezes the whole lane (healthy models get local 429/424) #5199 路径上的只解注释块,完整保留 429 governor hold is keyed by origin+credential with no model dimension — one model's quota exhaustion freezes the whole lane (healthy models get local 429/424) #5199 的 dualIdentity 字段 / 检测逻辑 / 告警日志代码,并把注释更新为 [Bug] compress can silently fold another message after restart when a pre-restart ref is reused (0.1.209) #5196 之后的准确表述。对 issue5199 新测试,改成手工构造 pre-[Bug] compress can silently fold another message after restart when a pre-restart ref is reused (0.1.209) #5196 持久化下来的双身份终态(byRef.m00006="f"、restampSuperseded={m3→b, m6→e, m7→f}),这样即使在 [Bug] compress can silently fold another message after restart when a pre-restart ref is reused (0.1.209) #5196 内核下,仍能真实覆盖宿主那条「检测到被复用的 ref 号就告警」的行为。

  4. 改了什么 / 推送:一个合并提交(master 基线同步)+ 一个独立标注的测试适配提交 test(#5199): build dual-identity fixture by hand …。远程快进推送成功(旧头 → 新头),未使用 force-push。

  5. 是否有回归问题:无。429 governor hold is keyed by origin+credential with no model dimension — one model's quota exhaustion freezes the whole lane (healthy models get local 429/424) #5199 的核心代码原样未动;两个整文件取 master 的文件本就与 429 governor hold is keyed by origin+credential with no model dimension — one model's quota exhaustion freezes the whole lane (healthy models get local 429/424) #5199 无关,收敛到 master 才是正确对齐;测试只是换了夹具的构造方式,所断言的宿主行为没有任何变化。

  6. 是否破坏缓存 / 是否改设计方向 / 是否引入新配置:均不涉及——纯冲突解决 + 测试夹具适配,没有运行时行为变更,没有缓存 / 重试 / 增量游标改动,也没有任何新配置项。

  7. 验证与当前状态:定向测试 8/8 通过(issue5199 T1–T4 + issue4797 T1–T4);npm run typecheck 干净(exit 0);推送后经 GitHub API 确认 mergeable=true(冲突已消除),mergeable_state=blocked(等待 CI / 评审,不是冲突)。全量回归按纪律交给远程 CI 把关,本地没有跑全量套件。

结论:冲突已解、红 CI 隐患已修、已推送,PR 现在可以进入 CI / 评审流程。

@ranxianglei

Copy link
Copy Markdown
Owner Author

🤖 Powered by ework · qwen3.8-27b

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

No new action needed this run — verifying live rather than trusting the earlier flag:

So the base-conflict from the previous trigger remains resolved and pushed; this forward reflects a snapshot taken before those landed. Nothing to fix or push.

状态:冲突已在上一次处理并推送(head=3de730d343);master 虽已前进到 c26c31ace(#6260),但该提交已包含在 PR head 内,未产生新冲突;GitHub 现为 mergeable=true / blocked(等 CI 与评审,非冲突)。本次仅做实时核验,无新增改动、无推送。

@ranxianglei

Copy link
Copy Markdown
Owner Author

🤖 Powered by ework · qwen3.8-27b

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

Re-verified live — nothing changed since my last two replies, so no action taken: remote PR head is still 3de730d343 (the already-pushed resolution), origin/master (c26c31ace) is fully contained in it (no new conflict), and GitHub reports mergeable: true / blocked (awaiting CI + review, not a base conflict). The base-conflict blocker from this forward is already cleared and pushed; nothing to fix or push this run.

@ranxianglei

Copy link
Copy Markdown
Owner Author

🤖 Powered by ework · qwen3.8-27b

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

Fourth forward of the same base-conflict blocker; live check again shows nothing changed — PR head 3de730d343 (already-pushed resolution), origin/master c26c31ace fully contained in it, GitHub authoritative mergeable: true / blocked. The conflict was resolved and pushed on the first run; these repeat forwards track a merge-state snapshot that has since cleared upstream, so no further agent action will alter the outcome. Nothing to fix or push.

@ranxianglei

Copy link
Copy Markdown
Owner Author

🤖 Powered by ework · qwen3.8-27b

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

PR #5212 独立复审结论 —— head 3de730d343(基线冲突解决后的新 head,非 floor 7/11 审过的旧 head 2fb797cf2)。 说明:floor 34 派专职 reviewer 复审,前两次(7/11)针对的是 #5196 kernel 合入前、尚未解冲突的 head;本轮我按当前 diff 从头重审,不复用旧结论。

八问:

  1. 是什么问题:给 429 governor hold is keyed by origin+credential with no model dimension — one model's quota exhaustion freezes the whole lane (healthy models get local 429/424) #5199 的「双身份 ref」补可见性。重启冷加载 ledger 重排后,一个 ref 号可能对当前持有者 live、同时又是另一条消息的 pre-restamp 记录标签;live-first 裁决下,引用该号的压缩调用被静默应用到新持有者并返回成功(块只覆盖错误消息、零日志)。本 PR 不改任何接受行为,只把这种原本静默的应用打一条醒目服务端日志。

  2. 什么原因导致的:根因在 kernel——restamp 同一次扫描里把空置号作为别的消息 seq+1 目标重新发放([Bug] compress can silently fold another message after restart when a pre-restart ref is reused (0.1.209) #5196 之前);[Bug] compress can silently fold another message after restart when a pre-restart ref is reused (0.1.209) #5196 slot-retirement 后新建会话不再产双身份号,但 [Bug] compress can silently fold another message after restart when a pre-restart ref is reused (0.1.209) #5196 之前 restamped 的旧会话仍带着该持久化形状。host 侧 live-first 本身是接受的裁决(After the agent restarts, this session can no longer be compressed: the ref suggested by acp_status is flagged as "renum | 代理重启后该会话再也压不了缩:acp_status 建议的 ref 被 compress 判为「已重编号」,两面对 ref 的认知不一致 #4797:拒绝会死循环),不是病因。本 PR 明确是「无决策依赖的可见性部分」,根治方向(kernel 不复用已发放号 / prune 复活)等 decision request: restamp reuses already-issued refs — replayed pre-restart citations silently compress the wrong message (#5199) #5204 拍板。

  3. 影响范围与严重程度:残留仅「restamp 早于 [Bug] compress can silently fold another message after restart when a pre-restart ref is reused (0.1.209) #5196 + 模型重放 pre-restart 引用」的旧会话触发,触发时是静默折叠错误消息 + 零日志(p1 数据丢失类)。本 PR 不消除残留、只让其可见;改动本身零行为变更。

  4. 怎么修复的:接受/拒绝/翻译逻辑逐字节不动;supersededRefRejection 新增 dualIdentity: string[](Set 去重,fast-path 空数组),在 resolveEndpoint 的 live 分支检测「live 且是另一条消息的记录标签」(oldOwner !== undefined && oldOwner !== liveOwner)记入 Set;applyRanges 对非空 dualIdentity 打一条 salted-id 告警(fix: replay connect-phase transport failures and report committed compression when the re-request dies (#1453) #1454 模式,只出 ref + salted owner id,无业务内容)。第二个 commit 修两处与代码事实不符的注释(ref-keyed-surfaces.md 把已 revert 的 prune 写成 enforced、以及被 429 governor hold is keyed by origin+credential with no model dimension — one model's quota exhaustion freezes the whole lane (healthy models get local 429/424) #5199 证伪的「危害已消除」断言;issue4797 测试头注释描述的 post-prune 形状)。出处已核:[PR] fix(#4797): root-cause fix for superseded live-key collision — restamp pruning + semantics-preserving guard translation | fix(#4797): superseded 活键碰撞根治——restamp 剪枝 + 守卫保义翻译 + 一次性标志修复 (kernel 0.0.166) #4807 确为起草后 revert 的 kernel prune。

  5. 是否有回归问题:逐项过既有行为——(a) superseded-ref 拒绝路径:rejection 早返回在新日志之前,拒绝语义不变;(b) fast-path(无 restampSuperseded):字段存在为空数组(T3 钉住);(c) 正常 live 引用:不误告警(T2 负对照);(d) 翻译路径:genuinely-vacated cite 仍翻译到同一消息、不误报双身份(T4)。关键点已逐行核过:resolveEndpoint 的返回值(live 原号 vs 翻译目标)与改前逐字节一致,dualIdentity 只是旁路写进一个纯供日志用的 Set、不进任何 accept/reject/translate 决策;out.dualIdentity 两条出口(fast-path 初始化 L944 / 正常路径 L978)都必设,全仓仅此一处调用方构造该结构,无遗漏。docs commit 仅动 .md 与注释,无代码。另见下方一条非阻塞措辞观察。

  6. 是否破坏缓存:不涉及——缓存/重试/增量游标/去重键均未触碰,仅新增一条服务端日志行(且只在双身份真实发生时出现)。

  7. 是否改了设计方向:没有。live-first 维持;根因方向留给 decision request: restamp reuses already-issued refs — replayed pre-restart citations silently compress the wrong message (#5199) #5204,与本 PR 的分工声明一致。

  8. 是否引入了新的配置:没有。

非阻塞观察(措辞,不拦):告警落在 applyRanges L1244——已过 superseded-ref 拒绝(L1221)与 out-of-recommendation(L1240)两道门,但仍在 stub-resubmission(L1277)/coverage-regression(L1321)/no-net-shrink(L1382) 等后续门之前。极端双重罕见情形(同一次调用既带双身份 ref、又触发后续某道门拒绝)下,日志字面「compress applied onto…」但实际未 fold。诊断信号本身仍正确(确实引用了复用号),且后续门会各自再打拒绝行、全貌不丢;这是措辞层面、不影响正确性。是否把「applied」调成「applying」或移到 fold 提交点之后,由作者定夺——我不擅自改写作者有意的措辞,也不为此开新条目。

测试有效性(④):四例真实有牙——T1 钉住核心回归(折的是当前持有者 f 而非旧持有者 e)+ 告警触发且点名 m00006;T2 负对照(普通 live cite 不告警,满足 #3617 双向断言);T3 契约边界(singleton 去重 / clean 空 / no-map fast-path 空);T4 翻译路径 intact(vacated cite 仍翻译、不误报)。夹具手工构造 pre-#5196 持久化形状(byRef.m00006="f"、restampSuperseded={m3→b,m6→e,m7→f})而非走 restampRefsToLedger——因为 #5196 后该函数会把撞号新戳停在 m8、不再复用 m6,restamp 路径已无法复现夹具;理由在注释里讲清了,诚实且让宿主守卫 pin 独立于 kernel 当前算法。

CI:当前 head 3de730d343 镜像缓存 30 checks / 0 failed / 0 pending(17 success + 13 skipped,skipped 全部是需要 ACP_TEST_* gate 的真客户端 e2e 车道,按设计跳过)——typecheck/build/核心 test shard/guard lane 全绿。

判定:独立复审通过,无拦下项。 代码审核 + 回归清点 + 测试有效性均过;CI 绿。可进入合并流程,等 owner 审批。(顺带一提:floor 26 提到的「#5360/#5196 合入后 stream.ts docblock 需小改」已在本次冲突解决中一并落到位——docblock 现表述为「#5196 起不再新造双身份号,本日志只覆盖 pre-fix 旧会话 + displacement 残留」,与代码一致。)

@ranxianglei

Copy link
Copy Markdown
Owner Author

🤖 Powered by ework · qwen3.8-27b

[bot] 🏷 (ai: vllm-qwen/qwen3.8-27b)

复审已完成并验证。完整结构化八问在上一条(floor 44),此处给一份自包含结论,不重复整段。

做了什么:对 PR #5212 当前 head 3de730d343(基线冲突解决后的新 head)做独立复审——按当前 diff 从头重审,不复用 floor 7/11 针对旧 head 2fb797cf2 的结论(那两个 head 是 #5196 kernel 合入前、未解冲突的)。

已验证(证据):

结论:独立复审通过,无拦下项。 代码审核 + 回归清点 + 测试有效性均过,CI 绿,可进入合并流程。

剩余与风险:仅剩 owner 合并审批(合并权在 owner,agent 不代合)。无阻塞风险;一条非阻塞措辞观察(告警字面「applied」位于后续若干拒绝门之前,极端双重罕见情形下字面略超前,但诊断信号仍正确)已在 floor 44 详述,交由作者定夺,不拦合并。

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 p1 scope/kernel

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant