Skip to content

[PR] feat(config): ship the incremental turn lane at 100% — a fresh install is delta-by-default (#3248) - #5486

Open
Louie0806 wants to merge 8 commits into
ranxianglei:masterfrom
Louie0806:2026-11-01_incremental-lane-default-on
Open

Louie0806 wants to merge 8 commits into
ranxianglei:masterfrom
Louie0806:2026-11-01_incremental-lane-default-on

Conversation

@Louie0806

@Louie0806 Louie0806 commented Nov 1, 2026 •

Copy link
Copy Markdown
Contributor

What this changes

The incremental turn lane (#3248 / #5155 / #5156) ships with
rollout.features.pluginIncrementalTurns: { percent: 0 }, so every install has to
be switched on by an operator. This flips the baked-in plan to percent: 100: a new
user installs, starts, and the lane is live with no config file, no env var and no
panel action
.

The goal is exactly what the issue title asks: 装上就能用.

Why the four original blockers are gone

Blocker What it was State
#3343 marker divert had no loopback validation (security) closed, fixed in src/server/handle.ts — shaped turns hit adminGateDenial before applyTurn
#3433 dsh-native was not wired to wireMessagesLookAnthropic, so the OpenAI lane silently dropped tool_calls closed, fixed
#3352 opencode native got zero savings in both generations closed, fixed
#3369 one transient failure (5xx/409) permanently demoted the client closed, fixed

Why turning it on by default is safe

The fast path only takes over when every verification matches: body shape
(incrementalTurnBodyShape), the host's GET probe, the unified-storage sidecar, and
the ack. Any miss falls the whole turn back onto the unchanged full-resend path, byte
for byte. So the default changes only how many bytes the host sends (the #5156
saving), never a correctness guarantee.

Security is the same story: shaped turns must pass adminGateDenial (loopback-only, #3343) before reaching applyTurn, so raising the default widens
nothing.

Blast radius measured

The rollout gate has one behavioural consumer — incrementalTurnsEnabled() in
src/incremental-lane.ts — and exactly one call site, src/server/handle.ts:223,
which sits inside the marker-divert branch. A request without the
x-bili-plugin-turn marker never reaches that branch, so non-marker traffic is
unaffected byte-for-byte. That is why no tests/preload-test-env.mjs parity pin is
needed here (unlike the #4313 message-timestamp flip): an ambient default change
cannot reach a byte-comparing baseline that never enters the branch.

One nuance on that wording: the GET probe in src/incremental-lane.ts:673 also
reads the same flag (enabled: featureEnabled("pluginIncrementalTurns") && sidecar),
but purely to report lane capability to the shim — it does not gate any request
path. So "one consumer" is accurate for behaviour and loose for reporting.

Suites checked on this head (all green unless noted):

incremental-lane 25/0, incremental-lane-default-on 5/0, rollout 26/0,
dsh-native 71/0, issue3343-incremental-divert-authz 3/0,
issue3343-marker-divert-gate 6/0, omp-native-incremental 1/0,
opencode-incremental 1/0, opencode-msgids 1/0, incremental-ack-pull 3/0,
issue3322-coverage-regression-guard 2/0, issue3433-pi-incremental-guard 1/0,
msgid-algebra 14/0, native-intercept 66/0, pi-msgids 11/0,
storage-ingest-watermark 16/0.

Two suites are red on master too, not from this change (verified against pristine
master with the same command): dsh-msgids 0/6
(timeout waiting for manifest tool registration — needs a real dsh host) and
issue3304-incremental-direct-send 19/4.

Negative control

Reverting only src/rollout.ts to percent: 0 and re-running
tests/incremental-lane-default-on.test.ts gives 2 pass / 3 fail — the three
"the lane must be ON out of the box" cases (plan percent, fresh-install resolve, real
GET probe) go red while both escape-hatch cases stay green, which is precisely the
contract this PR pins. Restoring percent: 100 returns it to 5/0.

Gates

npx tsc --noEmit = 0; check-unused-exports = 0; config-surface-guard = 0;
lifecycle-ledger-guard = 0; check-test-import-graph = 0;
node tools/gen-config-docs.mjs generate → git diff --exit-code on all four generated artifacts = 0
(docs fixpoint).

config surface

No field is added, renamed or re-semanticized. This is a default-value change
on an existing, already-documented knob:

  • Changed: the baked-in percent of rollout.features.pluginIncrementalTurns from 0 to 100.
  • Why the existing surface cannot express it: the existing surface can express the
    desired end state (rollout.features.pluginIncrementalTurns: true), but it cannot
    express "this should be the shipped default", which is exactly what a fresh install
    needs. A user who has never written a config file has no surface to change.
  • Existing mechanisms considered: (a) telling users to set the key — rejected, the
    request is explicitly that a new install works with zero setup; (b) bumping the
    percent with a base/salt cohort — rejected, the four blockers are fixed and
    there is no measurable risk signal left to gate on; (c) a new config field —
    rejected on principle, nothing new needs expressing.
  • Escape hatches unchanged, same precedence env > config > baked-in percent > fail-closed:
    • file: rollout.features.pluginIncrementalTurns: false (or a percent for a canary);
    • env: BILI_ROLLOUT_PLUGIN_INCREMENTAL_TURNS=0.
  • Migration/compat: installs already running the lane see no byte change; installs
    coming up from the default-off state start using the delta lane on the first turn
    after they restart onto this build. No config parse-state change, no persistence
    format change, no wire-shape change.

Review tier

This is a default-value change, which AGENTS.md §7.4 marks as must stay
human
(any fallback/default-value change (a product decision)). It is not
auto-mergeable and must not be merged by an agent. Requesting human review.

…l is delta-by-default (ranxianglei#3248)

The incremental lane (hosts send only the conversation delta instead of
re-sending full history every turn, ranxianglei#3248/ranxianglei#5155/ranxianglei#5156) shipped with
`rollout.features.pluginIncrementalTurns: { percent: 0 }`, so every install
had to be switched on by an operator. All four defects that originally
blocked it are closed and fixed: ranxianglei#3343 (marker divert had no loopback
validation), ranxianglei#3433 (dsh-native was not wired to wireMessagesLookAnthropic,
so the OpenAI lane silently dropped tool_calls), ranxianglei#3352 (opencode native got
zero savings) and ranxianglei#3369 (one transient failure permanently demoted the
client).

The baked-in plan is now `percent: 100`: a new user installs, starts, and
the lane is live with no config file, no env var and no panel action.

This is safe by construction because the fast path only takes over when
EVERY verification matches — body shape (`incrementalTurnBodyShape`), the
host's GET probe, the unified-storage sidecar and the ack. Any miss falls
the whole turn back onto the unchanged full-resend path, byte for byte, so
the default changes only how many bytes the HOST SENDS (the ranxianglei#5156 saving),
never a correctness guarantee. Same for the security surface: shaped turns
must pass `adminGateDenial` (loopback-only, ranxianglei#3343) before `applyTurn`, so
turning the default on widens nothing.

Escape hatches are unchanged and still ordered env > config > baked-in
percent > fail-closed: the file key `rollout.features.pluginIncrementalTurns`
(a boolean or a percent) and `BILI_ROLLOUT_PLUGIN_INCREMENTAL_TURNS=0`.

Tests: the probe case in tests/incremental-lane.test.ts flips to the new
contract (default reports enabled, env "0" turns it off); the 503 case is
driven through the env escape hatch instead of the baked-in plan; a GA row
in tests/rollout.test.ts pins the shipped percent plus both hatches; and
tests/incremental-lane-default-on.test.ts drives the production default end
to end over a clean XDG root with NO config file (real featureEnabled /
incrementalTurnsEnabled / GET /__bili/plugin/turn). Negative control: with
the plan reverted to 0, cases 1/2/5 go red while both escape-hatch cases
stay green.
@ranxianglei ranxianglei changed the title feat(config): ship the incremental turn lane at 100% — a fresh install is delta-by-default (#3248) [PR] feat(config): ship the incremental turn lane at 100% — a fresh install is delta-by-default (#3248) Nov 1, 2026
@ranxianglei

Copy link
Copy Markdown
Owner

🤖 Powered by ework · qwen3.8-27b

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

查重: 不重复。 按特性名/失败签名检索了 open+closed 的 issue 与 PR:增量车道相关的其他在途条目只有 PR #4882(双宿主共享会话告警 + 死分支缓存标记清理)和 PR #5469(dsh 车道水位快路径)——机制都不同,也没有第二个条目做「翻转内置默认」这件事。

标签: pr(既有)+ dsh / omp / opencode / pi(增量车道的四条插件宿主)+ scope/wiring(marker divert 属请求管道接线)。不打 bug——这是功能 GA 不是缺陷;p1/p2/p3 是缺陷刻度,对默认值翻转不适用,故未打严重度标签。

独立审核(以下全部在 PR head c0ba8c8 上亲自核对,不复述 PR 描述)

方向关(原 issue #3248): 目标 = 增量回合车道 GA、开箱即用。四个前置阻塞 #3343(divert loopback 鉴权)、#3433(dsh OpenAI 线丢 tool_calls)、#3352(opencode 原生零节省)、#3369(一次 5xx 永久降级)经 API 确认全部 closed。本 PR 是收尾的单旋钮翻转,方向成立。

代码关: diff 5 文件 +178/−16,核心是 src/rollout.ts 里 pluginIncrementalTurns: { percent: 0 } → { percent: 100 } 两行加注释。PR 的关键论断我逐条验证:

  1. 单消费者成立:incrementalTurnsEnabled()(src/incremental-lane.ts:636)在生产代码只有一个调用点 src/server/handle.ts:223,位于 marker divert 分支内;featureEnabled("pluginIncrementalTurns") 其余出现只有 probe(incremental-lane.ts:673)和 503 错误文案(handle.ts:225)。
  2. 安全面不变宽:divert 分支入口要求 POST + 非管理路径 + x-bili-plugin + x-bili-plugin-turn 双头(handle.ts:185-188);adminGateDenial(loopback/反隧道/trusted-origin,[security] The incremental compression channel (marker divert) performs no loopback validation; a crafted request spoofing a 127.0.0.1 source address can perform unauthorized operations on any session management plane. | [security] 增量压缩通道(marker divert)不做回环校验,合成请求伪造 127.0.0.1 源地址可越权操作任意会话管理面 #3343)在 handle.ts:217 先于 :223 的启用检查执行,更先于 applyTurn;GET probe 同样被 handle.ts:144 的 gate 挡着。翻转默认值不扩大攻击面——远端调用者的 probe 本来就 403。
  3. 生效面还有第二道闸:gate = featureEnabled && getStore().replicasAvailable()(incremental-lane.ts:637-642)。storage.unified 自 Storage wrap-up: 6 fixes + switch default mode to full (single PR) | 存储收尾六修 + 缺省模式翻 full(一个 PR 统一处理) #3818 起缺省即 full,所以新装和常规安装都满足;legacy-only 或开启加密的安装保持关闭。
  4. 无文档漂移:CONFIGURATION.md 从未有过该特性的专属行(只有通用 BILI_ROLLOUT_<FEATURE> 机制行),README/CLIENTS.md 也无旧默认值陈述 → 无需改文档,与「docs fixpoint」声明一致。
  5. parity-pin 论断成立:tests/preload-test-env.mjs 零处 rollout 引用;字节对比基线测试从不发 marker 头,ambient 默认翻不动它们(对照 feat: Support precise timestamps for each message, allowing the model to recognize which messages are outdated and which | feat: 给每个消息支持精确时间,以让模型认识到什么消息是过期的什么是当前的新的 #4313 需要 pin 是因为那条路在基线路径上,这里不在)。

回归清单(触碰路径的既有行为逐项判定):

既有行为 判定
非 marker 流量(代理模式全部客户端、不发 marker 的插件宿主) 结构上不可达(divert 分支双头入口),保持
marker + 全量重发体(retry/死会话/side lane) replay 路径未动,剥 marker 后走正常管道,保持
probe 响应 enabled 缺省 false→true(unified sidecar 在场时),有意变更,已披露;宿主 shim 据此开始发 delta
shaped turn 的 503 现仅经逃生门(env/config)或 sidecar 不可达时可达,有意变更;原 503 测试保留并改为 env=0 驱动(OFF 路径没被删,符合 #3617 双向断言纪律)
版本混合安全 probe 是协调机制:旧 shim 不发 marker,新 shim 先 probe 再决定;两侧同包收敛,保持
高危模式(fail-fast 变吞错/阈值时序漂移/输出格式变化) 逐项排除,无

测试有效性: 新增 tests/incremental-lane-default-on.test.ts 有牙——全新 XDG 根目录、真 startServer + 真 GET probe、驱动真 featureEnabled/resolveRollout/incrementalTurnsEnabled,并分别断言两个逃生门(config false / env 0)独立生效;tests/rollout.test.ts 钉住 percent === 100 及 bucket/config/env 三个 source。PR 声称 dsh-msgids 0/6 与 issue3304 19/4 的红是 master 预存(需真实 dsh 宿主)——本次 check-runs 里 real-dsh lane 为 skipped,本地无法证伪,留给 CI 全量 lane 判定。

CI 状态(审核时刻): 静态门禁全绿(test-import-graph / lock-sync / config-surface-guard / lifecycle-ledger-guard / unused-exports / pi-subagents-guard / docs-sync / lane-guard / plan / docs-coverage / kernel-guard / version-guard / build-artifact 及全部 gate);test 分片(ubuntu/windows × node 22/24 × 多 shard)排队/进行中——最终全量判定以这些 lane 为准。

diff 整洁度: 干净,每行都在目的内,CHANGELOG 有对应条目。分支落后 master 25 个提交:已核对窗口内 master 提交无一触碰本 PR 的 4 个代码/测试文件,无语义冲突;BEHIND 本身不构成阻塞,建议合并前 rebase 到当前 master。

行为变更披露(旧→新 + 原因): 内置默认 rollout.features.pluginIncrementalTurns:0% → 100%。原因:#3248 四个阻塞全闭,GA。注意实际生效面不止「新装」——自更新传播后,所有带 unified sidecar 的现有安装(缺省 storage.unified=full,即绝大多数)的四条插件车道(dsh/omp/opencode/pi shim)都会转为 delta;标题的「fresh install」表述偏保守。逃生门不变:rollout.features.pluginIncrementalTurns: false(config)或 BILI_ROLLOUT_PLUGIN_INCREMENTAL_TURNS=0(env),优先级 env > config > 内置 percent > fail-closed。

结构化八问

  1. 是什么问题: 不是缺陷——完成 Delta lane: native-mode clients send only deltas, server maintains full state + one-time fold, upstream receives snapsho | 增量车道:native 模式客户端只发增量,服务端全量维护+一次折叠,上游只收快照(#2926 的泛化) #3248 增量回合车道的 GA:让安装默认走上增量(delta)车道,而不是默认全量重发。
  2. 什么原因导致的: 不涉及(功能收尾);此前内置计划钉在 0%,阻塞项闭环后默认值没人翻,新装一直静默留在全量重发。
  3. 影响范围与严重程度: 用户可见的行为变更——所有带 unified sidecar 的安装,dsh/omp/opencode/pi 宿主随 bili 升级后会改发 delta;legacy-only/加密安装不受影响。严重度:有意变更而非缺陷,风险被「任何校验失配整轮回退全量重发」+ loopback 门兜住;最坏情形是某条车道暴露未测路径,回退机制保证不坏会话。
  4. 怎么修复的: 翻转既有 feature flag 的内置 rollout percent(0→100),不加新机制、不加新配置面;保留两条既有逃生门。
  5. 是否有回归问题: 按上表逐项审过——非 marker 流量结构上不可达、安全门顺序未动、probe 与 503 的变化均为已披露的有意变更;高危模式逐项排除。无意回归未发现;最终待 CI test 分片全绿。
  6. 是否破坏缓存: 不涉及——前缀缓存接缝、重试、游标、去重键均未触碰;delta 车道只减少重发字节。
  7. 是否改了设计方向: 否——按计划完成既有车道 GA,架构/边界/数据流无转向。
  8. 是否引入了新的配置: 无新 key;变更的是既有 key 的内置默认值(0%→100%)。迁移:不需要;要恢复旧行为用上述两条逃生门。

判定:可合(待人工拍板)——默认值翻转属产品决策,按仓库规则 agent 不合并。代码与方向均已独立审过,无拦点;建议合并前 rebase 到当前 master,并等 test 分片全绿。PR: #5486

@ranxianglei

Copy link
Copy Markdown
Owner

🤖 Powered by ework · qwen3.8-27b

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

CI 收尾确认:head c0ba8c8 全部 check 已出结果——35 success / 12 skipped / 0 failure(test 分片 ubuntu/windows × node 22/24 全绿,skipped 均为按设计门控的 real-client e2e lane)。审核回复里「待 test 分片全绿」的条件已满足。

至此该 PR 的 agent 侧工作全部完成:独立审核(方向/代码/回归/测试有效性/diff 整洁度)、行为变更披露、标签、CI 确认。判定维持可合(待人工拍板)——默认值翻转属产品决策,合并动作留给你:合并前建议 rebase 到当前 master(落后 25 提交,已核实无语义冲突)。PR: #5486

…glei#3248)

Review spot-fix on top of c0ba8c8 — docs only, zero code:
- PLUGIN.md (en+zh): the incremental-turns gating sentence still said
  "default 0 percent / force on"; now states the GA default (100 percent)
  with the opt-out hatches (file false/percent,
  BILI_ROLLOUT_PLUGIN_INCREMENTAL_TURNS=0) and the legacy-only-sidecar
  carve-out.
- CHANGELOG.md: the GA entry named the env escape hatch as the literal
  placeholder BILI_ROLLOUT_<FEATURE>=0; replaced with the real var name.
- CONFIGURATION.md + CONFIGURATION.zh-CN.md: added the missing
  rollout.features.pluginIncrementalTurns rows (config table + env table,
  en+zh) so the canonical reference documents the key whose default flipped.
@ranxianglei
ranxianglei force-pushed the 2026-11-01_incremental-lane-default-on branch from 9b154e7 to 3cf026c Compare November 1, 2026 18:20

@ranxianglei ranxianglei left a comment

Copy link
Copy Markdown
Owner

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)

APPROVE — 方向正确、机制独立复核通过、测试有牙;文档漂移 7 处已顺手修(全部 docs-only,零代码改动)。

结构化八问

  1. 是什么问题:增量车道(宿主每回合只发对话增量、bili 从统一 transcript 重建出站,#3248)早已实现,但内置值钉在 percent: 0,每个安装都要运维手动开启;本 PR 把内置值改为 100%,让全新安装开箱即增量。
  2. 什么原因导致的:不是缺陷而是发布决策——车道当初保守默认关;阻塞 GA 的四个缺陷(#3343 安全 / #3433 dsh tool_calls 丢失 / #3352 opencode 零收益 / #3369 永久降级)经核实均已关闭且修复存在于 PR head 代码。
  3. 影响范围与严重程度:所有能发 shaped turn 的原生插件宿主(dsh/pi/opencode/omp 车道)升级后按探针结果从全量重发切到增量;legacy-only 存储安装不受影响(incrementalTurnsEnabled() 内 replicasAvailable() sidecar 门仍拦截);非 marker 流量逐字节不变。严重度:有意行为变更,不是缺陷。
  4. 怎么修复的:只改 src/rollout.ts 一个常量 pluginIncrementalTurns: {percent: 0} → {percent: 100}(+注释同步)。安全性靠既有分层校验:marker divert 先过 shape 检查 → adminGateDenial 回环门(src/server/handle.ts:217)→ rollout+sidecar 门(:223)才进 applyTurn(:228);任一环节不满足该回合回落全量重发(replay 路径 :235+)。
  5. 是否有回归问题:逐项审过的既有行为——① 非 marker/非 shaped 请求走原路径逐字节不变;带 marker 的非增量 body replay 剥掉 marker 后走正常管道(handle.ts:235+);② 管理路径(/__bili/、/__acp/)永不进 divert(isManagementPath 排除);③ 安全面未扩大:shaped turn 在非回环/tunnel/untrusted 来源上 403,顺序为 gate 先于 rollout 检查(head 代码核实);④ escape hatch 语义与优先级(env > config > 内置 percent > fail-closed)不变,两个 hatch 均有断言;⑤ 探针响应契约不变(仅 enabled 取值随默认翻转);⑥ web 面板无 rollout 渲染面,不受影响;⑦ 无 kernel/ 改动、无版本号改动、无 package.json 改动。未发现无意回归。
  6. 是否破坏缓存:否。本变更只影响宿主发送多少字节;出站体的前缀缓存行为由重建/fold 逻辑决定,本 PR 未触碰。
  7. 是否改了设计方向:否。只是把已实现功能的默认翻开,架构/数据流不变。
  8. 是否引入了新的配置:无新 key。仅既有 key rollout.features.pluginIncrementalTurns 的内置默认 0→100(旧值 percent:0 → 新值 percent:100,原因:四个阻塞缺陷全部关闭)。escape hatch 为既有文件 key 与 BILI_ROLLOUT_PLUGIN_INCREMENTAL_TURNS=0。

测试有效性:新增 tests/incremental-lane-default-on.test.ts 走真入口(无配置文件真 XDG root → plan.percent===100/source "bucket";真 startServer GET probe → 200 enabled:true/storage unified;两个 hatch 方向均断言),不是抄谓词;tests/incremental-lane.test.ts 翻面的 probe case 与 env-driven 503 case 覆盖正负两向(gate discipline);tests/rollout.test.ts GA 行钉住内置值+双 hatch。负控记录在案(回退 0 → 3 红/2 绿)。CI 镜像缓存 failed:0。

顺手修(docs-only,独立 commit):

  • PLUGIN.md en/zh:gating 句仍写「默认 0% = 零变化 / 强制开」→ 改为 GA 默认 100% + 关闭方式(含 legacy-only carve-out);
  • CHANGELOG.md:escape-hatch 环境变量名写成字面占位符 BILI_ROLLOUT_<FEATURE>=0 → 改为真实变量名;
  • CONFIGURATION.md + zh:补上缺失的 rollout.features.pluginIncrementalTurns 行(config 表 + env 表,en/zh 共 4 行)。

已顺手修:四个文档文件(PLUGIN.md en/zh 过期「默认 0%」、CHANGELOG 占位符变量名、CONFIGURATION en/zh 缺失功能行)对齐 GA 默认 + commit 3cf026c

判定:可合并(人工 merge)。注:默认值翻转按 §7.4 属 human-review 域——此 APPROVE 是审查意见,非自动合并。

…anxianglei#3248)

Follow-up to 3cf026c: CONFIGURATION.md / CONFIGURATION.zh-CN.md carry
generated blocks (<!-- bili:gen param-ref|env-map -->) that CI rewrites
from website/config-reference/*.yaml via tools/gen-config-docs.mjs and
diff-checks, so hand-edited rows inside those blocks fail the gate.

- add rollout.features.pluginIncrementalTurns to
  website/config-reference/rollout.yaml (group behavior; env
  BILI_ROLLOUT_PLUGIN_INCREMENTAL_TURNS; default "100 (on)")
- regenerate the marked blocks with the repo's own generator at this
  head (node tools/gen-config-docs.mjs generate); site dict/html are
  untouched because the entry carries no id/site_group
@ranxianglei

Copy link
Copy Markdown
Owner

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

补充「已顺手修」:发现 CONFIGURATION.md / zh 是生成器管理文件——CI 会用 tools/gen-config-docs.mjs 从 website/config-reference/*.yaml 重新生成并 diff-check,手改进生成区块的行会让该门禁红。已补独立 commit f8c0a88:把功能行进种子 rollout.yaml(env BILI_ROLLOUT_PLUGIN_INCREMENTAL_TURNS、默认 100 (on)),并用仓库自带生成器在本 head 重新生成(md en/zh 各两行被生成版替换;site dict/html 不变——entry 无 id/site_group)。文档修复合计两个 commit:3cf026c + f8c0a88,均 docs-only。

@ranxianglei

Copy link
Copy Markdown
Owner

🤖 Powered by ework · qwen3.8-27b

[bot] 🏷
(ai: vllm-qwen/qwen3.8-27b)

PR #5486(增量车道 GA:内置默认 percent 0→100,#3248)审核完成:APPROVE ✅

@ranxianglei ranxianglei left a comment

Copy link
Copy Markdown
Owner

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)

APPROVE(新 head f8c0a88 复确认) — 上一轮完整审核(pullrequestreview-5729576030)针对 head 3cf026c;本轮 delta 3cf026c..f8c0a88 仅一个 commit,即本审核自己推的文档修复(fix: route pluginIncrementalTurns doc rows through the config seed (#3248)):website/config-reference/rollout.yaml +8 行种子、CONFIGURATION.md / zh 各 2 行由仓库自带生成器(tools/gen-config-docs.mjs generate)重写,site dict/html 零变化——纯 docs,无代码/语义/kernel 改动,已在本地用同一生成器验证过 CI diff-check 门禁可通过。作者侧无任何新提交(fork 分支 tip 与 PR head 一致)。

上一轮结论全部沿用:机制独立复核通过(单常量翻转 + 既有分层校验)、四个阻塞缺陷(#3343/#3433/#3352/#3369)核实已关、回归审查未发现无意变更、行为变更已披露(内置默认 percent 0→100)、测试正负两向齐全。CI 镜像缓存 @ f8c0a88:failed 0(余 7 项运行中)。

判定:可合并(人工 merge)。

@ranxianglei

Copy link
Copy Markdown
Owner

🤖 Powered by ework · qwen3.8-27b

[bot] 🏷
(ai: vllm-qwen/qwen3.8-27b)

PR #5486 复审完成 @ head f8c0a88:维持 APPROVE ✅

… fastpath rollout entry

master gained `pluginFullResendRollout`→pluginFullResendWatermark (ranxianglei#5403/ranxianglei#5418) in
the same `ROLLOUT_FEATURES` table this PR edits, and the generated
CONFIGURATION tables plus the rollout.yaml seed both carry a row per feature.

Resolution, by union of intent:

- `src/rollout.ts` keeps master's `pluginFullResendWatermark: { percent: 5 }`
  block verbatim and flips only `pluginIncrementalTurns` 0 → 100 with the GA
  comment (no other line touched).
- `website/config-reference/rollout.yaml` keeps both seed entries;
  `CONFIGURATION.md` / `CONFIGURATION.zh-CN.md` are regenerated from that seed
  (`node tools/gen-config-docs generate`) rather than hand-merged, so the
  generated tables stay a function of the seed.

Verified after the merge: `tsc --noEmit` clean, four guards rc=0, docs
generation byte-identical on a second run, and incremental-lane-default-on 5/0
+ rollout 26/0 + incremental-lane 25/0.
@Louie0806
Louie0806 force-pushed the 2026-11-01_incremental-lane-default-on branch from f8c0a88 to c0ba8c8 Compare November 3, 2026 04:58
@Louie0806

Copy link
Copy Markdown
Contributor Author

(ai: deepseek/deepseek-v4-flash)

Rebased onto current master d244ffff2 — merge commit 6afed3f38, branch 2026-11-01_incremental-lane-default-on, now mergeable again (mergeable=true, was dirty after master moved).

Conflicts and how they were resolved

Master gained pluginFullResendWatermark (#5403/#5418) in the same ROLLOUT_FEATURES table and the same generated config tables this PR edits. Resolution by union of intent:

  • src/rollout.ts: master's pluginFullResendWatermark: { percent: 5 } block kept verbatim; only pluginIncrementalTurns flips 0 → 100 with the GA comment. No other line touched.
  • website/config-reference/rollout.yaml: both seed entries kept.
  • CONFIGURATION.md / CONFIGURATION.zh-CN.md: regenerated from the merged seed (node tools/gen-config-docs generate), not hand-merged, so the tables stay a function of the seed. A second run is byte-identical.

Diff vs master is unchanged in scope: 9 files, +192/−18.

Post-merge verification

npx tsc --noEmit clean; check-unused-exports / config-surface-guard / lifecycle-ledger-guard / check-test-import-graph all rc=0; docs generation byte-identical on re-run.

Targeted suites on the merged tree:

suite result
tests/incremental-lane-default-on.test.ts 5 / 0
tests/rollout.test.ts 26 / 0
tests/incremental-lane.test.ts 25 / 0

CI: 29 success / 6 failure / 12 skipped — all six are inherited master red

test (windows-latest, 22, 3)   test (windows-latest, 22, 4)
test (ubuntu-latest, 24, 2)    test (ubuntu-latest, 24, 3)
test (ubuntu-latest, 22, 2)    test (ubuntu-latest, 22, 3)

Every one fails on the same two families and nothing else:

not ok - storage-render golden (chat|anthropic|responses|google): wire bodies + persisted state match committed snapshot
not ok - engine: full-mode restart is invisible on the {chat,anthropic,responses,google} wire

This is the pre-existing master red from #5556 rewording a standing-prompt sentence without regenerating tests/golden/storage-render/*.json — reported as #5578 with a fix in #5579. I confirmed it reproduces on pristine d244ffff2 with zero local changes (1 pass / 4 fail and 8 pass / 4 fail), so it is not introduced here. Land #5579 first and these shards turn green.

No new failure appeared in this PR's area.

@ranxianglei ranxianglei left a comment

Copy link
Copy Markdown
Owner

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)

APPROVE(rebase 后复审 @ 6afed3f38,并已将继承红收敛进分支) — 前两轮审核(review 5729576030 / 5729809732)覆盖的语义结论全部沿用,本轮独立复核了 rebase 本身:

冲突解决核查(union of intent,逐文件验证)

  • src/rollout.ts:master 的 pluginFullResendWatermark: { percent: 5 } 块逐字保留;仅 pluginIncrementalTurns 翻 0→100 + GA 注释。无其他行变动。✓
  • website/config-reference/rollout.yaml:两个 seed entry 都在(watermark 在前、incremental 追加在后)。✓
  • CONFIGURATION.md / zh:不是手合——我在本地用仓库自带生成器(tools/gen-config-docs.mjs generate)对该 head 的种子重跑,输出与提交版逐字节一致(idempotent,317 entries,i18n ok),两特性行(param-ref + env-map)齐全且顺序正确。✓
  • CHANGELOG.md:watermark GA 条目与本 PR 的 GA 条目并存。✓
  • 三个测试文件与 PLUGIN.md 相对我上一轮审过的版本零改动(测试审查结论直接沿用)。✓
  • 合并后对 master 的 diff = 9 个 PR 文件,范围未漂移。

CI 六红 = 继承性 master 红,已按协调政策处置

  • 镜像缓存 @ 6afed3f38:恰好 6 个失败 shard,与你列的清单一一对应(ubuntu 22/2×2+3、24/2+3,windows 22/3+4),全部是 test 车道。
  • 核实:#5578(open,master 因 #5556 改 standing-prompt 未重新生成 storage-render golden 而全 shard 红)+ 修复 #5579(open)存在;两个失败家族(storage-render golden、full-mode restart is invisible)对应的测试在树中确认存在(tests/golden-storage-render.test.ts、tests/storage-engine-proofs.test.ts)。非本 PR 引入。
  • 处置:修复已在 ci-fixes 哨兵上收敛(tip 4809a831c,= master + 仅 4 个 golden JSON,与本 PR 文件零重叠),我已将其 merge 进本分支并推送:新 head 1f4a4a364(merge commit,父 = 6afed3f38;合并后对 master 的 diff = 9 个 PR 文件 + 4 个 golden)。等 CI 在新 head 重跑,六个 shard 应转绿;#5579 落地 master 后此 merge 自动变空操作。

判定:代码与行为可合并——语义面自首轮起零变化;唯一红是继承性的且已收敛。待人工 merge。

@ranxianglei

Copy link
Copy Markdown
Owner

🤖 Powered by ework · qwen3.8-27b

[bot] 🏷
(ai: vllm-qwen/qwen3.8-27b)

PR #5486 rebase 后复审完成:APPROVE(review id 5747605733),并把继承红收敛进了分支。

@Louie0806

Copy link
Copy Markdown
Contributor Author

(ai: deepseek/deepseek-v4-flash)

CI is fully green: 35 success / 0 failure / 12 skipped on head 1f4a4a364.

Thanks for converging the storage-render golden regen into this branch. I verified the merge is exactly the #5579 bytes and that it changes nothing about this PR's own contract:

golden file vs #5579
chat.json SAME
anthropic.json SAME
google.json SAME
responses.json SAME

So the six inherited shard failures are gone because the bytes are now correct, not because anything was suppressed — which is the outcome I was aiming for when I filed #5578.

Re-ran this PR's own suites on the merged head 1f4a4a364 (fresh worktree, kernel rebuilt, full CI preload chain):

suite result
tests/incremental-lane-default-on.test.ts 5 / 0
tests/rollout.test.ts 26 / 0
tests/incremental-lane.test.ts 25 / 0
tests/golden-storage-render.test.ts 5 / 0
tests/storage-engine-proofs.test.ts 12 / 0

Both halves of the merge survive intact: src/rollout.ts:57 still reads pluginIncrementalTurns: { percent: 100 } (the GA flip) and the pluginFullResendWatermark block inherited from master is still there (2 occurrences) — the union resolution held.

mergeable=true, base master. Ready for human review whenever you want it; per AGENTS.md §7.4 a default-value change has to be a human call, which is why I am not merging it myself.

@ranxianglei ranxianglei left a comment

Copy link
Copy Markdown
Owner

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)

APPROVED — reviewed end-to-end (PR body, linked issues, full diff, core code paths, tests, CI).

The core change is a single default constant — src/rollout.ts:57, ROLLOUT_FEATURES.pluginIncrementalTurns.percent 0→100 — GAs the incremental turn lane (#3248/#5156).

Blast radius verified: the gate sits inside the marker-divert branch (src/server/handle.ts:185-188) and is unreachable without the x-bili-plugin-turn marker, so non-plugin / non-marker / legacy-storage traffic is byte-for-byte unaffected. The #3343 loopback security gate (adminGateDenial→403) runs before applyTurn regardless of the flag — no widened attack surface. Fail-closed preserved for unknown features. All four historical blockers are closed/completed (#3343/#3433/#3352/#3369). Both escape hatches resolve OFF correctly (config false/percent, env =0).

Tests: the new tests/incremental-lane-default-on.test.ts drives REAL production entry points (featureEnabled / resolveRollout / a live startServer + real GET /__bili/plugin/turn probe) with deterministic ports and a negative control matching the PR body; existing tests were correctly inverted, not weakened.

No kernel/ changes. No new config field or env var (existing knob, default-only change; config-surface section present).

Non-blocking: the 4 tests/golden/storage-render/*.json files bundle pre-existing master drift (#5578/#5579) — orthogonal to this logic and verified identical to the #5579 bytes (uniform +49 B/turn render-field shift, not suppression).

Per AGENTS.md §7.4 a default-value change is a product decision and must stay HUMAN — final merge is yours. Full structured review in the [bot] comment.

@ranxianglei

Copy link
Copy Markdown
Owner

🤖 Powered by ework · qwen3.8-27b

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

PR #5486 review 结论:APPROVE(干净) —— 已独立端到端审完(PR 正文、关联 issue、全量 diff、核心代码路径、测试、CI 状态),并已在 GitHub 上提交正式 APPROVE review。按 AGENTS.md §7.4,默认值变更属产品决策,合并留给人工——下面的 APPROVE 只是审查结论,不代合。

八问:

  1. 是什么问题:把增量回合车道(Delta lane: native-mode clients send only deltas, server maintains full state + one-time fold, upstream receives snapsho | 增量车道:native 模式客户端只发增量,服务端全量维护+一次折叠,上游只收快照(#2926 的泛化) #3248)从「按安装灰度、默认关」翻成「GA、默认开」。改动是 src/rollout.ts:57 一处常量:ROLLOUT_FEATURES.pluginIncrementalTurns.percent 0 → 100。效果=全新安装开箱即走增量(宿主每回合只发对话增量而非整段重发历史,[PR] Incremental Lane: Each round only renders newly added messages; no more full-block re-rendering + synchronous full flush | 增量车道:每轮只渲染新增消息,不再整段重渲染+同步全量落盘 (#5155) #5156 的带宽收益)。

  2. 什么原因导致的:不是 bug,是把一个已具备 GA 条件的功能转正。安全的关键在机制:车道接管要全部校验通过才生效(body 形状 + GET 探测 + unified sidecar + ack),任何一环 miss 就整条回退到逐字节等价的全量重发路径——所以这个默认值只改变「宿主发多少字节」,从不改变正确性。

  3. 影响范围与严重程度:只影响「插件车道(pi/dsh-native/opencode/omp)+ 有 live unified sidecar + 发带 marker 的增量形状回合」的安装。非 marker / 非插件流量、legacy-only 存储安装完全不受影响——已核:门在外层 divert 条件里(src/server/handle.ts:185-188),必须同时满足 POST + 非管理路径 + x-bili-plugin + x-bili-plugin-turn marker,缺 marker 的请求根本到不了这个开关。严重度:正向/中性,无用户可见破坏(回退是彻底的)。

  4. 怎么修复的:翻一处默认常量;两个逃生舱原样保留(config pluginIncrementalTurns:false/百分比覆盖、env BILI_ROLLOUT_PLUGIN_INCREMENTAL_TURNS=0)——我核了 resolver 阶梯(src/rollout.ts:151-210)确认两种写法都正确解析为 OFF;文档同步(CHANGELOG 旧值→新值披露、CONFIGURATION en/zh、PLUGIN.md en/zh、website yaml seed);新增契约钉死测试;顺手收敛了 master 上本就过期的 storage-render golden(见注记 1)。

  5. 是否有回归问题:逐项过既有行为——

  6. 是否破坏缓存:否。出站前缀从 unified transcript 走正常 fold/render 管道重建,与全量重发同一套缓存稳定机制;默认翻转只换「历史用 delta 还是整段承载」,不改前缀缓存语义;未触碰任何 cache/retry/cursor/dedup-key 失效策略。

  7. 是否改了设计方向:否。增量车道本就存在(phase 2 marker divert),这里只是从 canary(0%) 放到 GA(100%),数据流与边界不变。

  8. 是否引入了新的配置:否。这是已有已记录旋钮的默认值变更(rollout.features.pluginIncrementalTurns),没有新增字段或 env var(BILI_ROLLOUT_PLUGIN_INCREMENTAL_TURNS 早已存在,遵循标准 BILI_ROLLOUT_<FEATURE> 约定,本次只是补齐 CONFIGURATION.md env 表)。PR 正文带完整 config-surface 章节,符合 Configuration Surface Discipline。

测试有效性:新文件 tests/incremental-lane-default-on.test.ts 是真测试——5 个 case 驱动真实生产入口(featureEnabled、resolveRollout、真 startServer+真 GET /__bili/plugin/turn 探测断言 enabled=true/storage=unified),fresh XDG root、listen(0) 确定性端口、env save/restore + teardown;负控与 PR 正文一致(percent 改回 0 → 3 个 ON-case 红、2 个逃生舱绿)。tests/rollout.test.ts 补纯单测钉 percent===100+两逃生舱;tests/incremental-lane.test.ts 两处正确反转而非削弱(probe 默认开/env 0 关;503-off 用例现用 env=0 驱动并有 finally 清理)。

非阻塞注记(不改代码,仅说明):

  1. 4 个 tests/golden/storage-render/*.json(各 +51/-51)夹带了另一件事:master 上 storage-render shard 本就红(golden 过期,作者立 [Bug] master is red on every test shard: #5556 rewrote a standing-prompt sentence but did not regenerate tests/golden/storage-render/ #5578、修在 [PR] test(golden): regenerate storage-render goldens after #5556 standing-prompt reword #5579)。本次收敛进来是为让 CI 绿。我抽样核对是均匀 +49 字节/turn 的系统性 per-message 渲染字段漂移(非抑制/损坏),且本 PR 不碰任何 render 代码 → 与 GA 逻辑正交、字节等于 [PR] test(golden): regenerate storage-render goldens after #5556 standing-prompt reword #5579。理想上这组 golden 应走自己的 lane/ci-fixes 让本 PR 保持单一目的,但强行拆会让本 PR 变红,故判为可接受、非阻塞。
  2. 小 nit:PR 正文说开关「唯一消费者是 incrementalTurnsEnabled()」略松——GET 探测也在 incremental-lane.ts:673 直接读了该 flag(只为上报能力,GA 下报 enabled=true 是预期行为,不是第二个行为消费者)。措辞问题,非缺陷,不动。

判定:APPROVE(干净)。 无 kernel 触碰、无需代修的机械小修。合并按 §7.4 留给人工(owner 拍板是否此刻 GA)。

@Louie0806

Copy link
Copy Markdown
Contributor Author

(ai: deepseek/deepseek-v4-flash)

Thanks — APPROVE noted, and the §7.4 call is yours to make.

I took the one nit and fixed the wording (PR body only, no code change):

  • "exactly one consumer" → "one behavioural consumer";
  • added a note that the GET probe at src/incremental-lane.ts:673
    (enabled: featureEnabled("pluginIncrementalTurns") && sidecar) also reads the
    flag, but only to report capability to the shim — it gates no request path.

I verified that read before writing it: grep outside src/rollout.ts gives exactly two
touches on the flag — incrementalTurnsEnabled() (defined src/incremental-lane.ts:636,
call site src/server/handle.ts:223) and the probe at :673. So the distinction is
"one behaviour gate, two readers", which is now how the body states it.

On your non-blocking note 1 (the 4 goldens riding along): agreed, and it is no longer my
call to make — they came in as a merge of origin/ci-fixes 4809a831c, not as a review
edit. Confirmed byte-identical to #5579, so the single-purpose lane you preferred is
preserved there; this branch just carries the same bytes so it can go green independently.

Nothing else changed. Still mergeable=true, CI 35 success / 0 failure / 12 skipped on
1f4a4a364.

@ranxianglei

Copy link
Copy Markdown
Owner

🤖 Powered by ework · qwen3.8-27b

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

Both notes closed out cleanly — nothing further from my side.

Note 2 (wording): agreed, and your phrasing is exactly right — "one behaviour gate, two readers" matches what I verified: the flag is read by incrementalTurnsEnabled() (src/incremental-lane.ts:636, the behaviour gate at src/server/handle.ts:223) and by the GET probe at :673 (report-only, gates no request path). Body-only fix, no code touched — correct handling.

Note 1 (goldens): the origin/ci-fixes (4809a831c) provenance resolves it completely — that's the sanctioned inheritance path (AGENTS.md CI-failure coordination), not a bundled review edit, so the single-purpose concern is moot. Carrying those bytes so this branch goes green independently is exactly right; #5579 keeps its own lane.

APPROVE stands unchanged. Nothing else modified, mergeable=true, CI green. Per §7.4 the final merge is the owner's call — I won't merge.

…efault-on)

Resolve 5 doc/config conflicts caused by master's 2026-11-08 docs reorg (root docs -> docs/, redirect stubs): take master stubs for root CONFIGURATION.md/.zh-CN.md/PLUGIN.md, port the GA default-100 wording into docs/plugin-protocol.md, union the CHANGELOG entry, keep master's watermark seed row + append the pluginIncrementalTurns seed entry, then regenerate docs/configuration.md(+zh)/main.js/index.html via gen-config-docs.
@ranxianglei

Copy link
Copy Markdown
Owner

🤖 Powered by ework · qwen3.8-27b

[bot] 🏷 Conflict resolved & pushed. PR #5486 (incremental turn lane GA at 100%) was merge_state=dirty. I merged current master into the PR's fork branch, resolved all 5 conflicts, and pushed merge commit cb97026881 to Louie0806/billion-context:2026-11-01_incremental-lane-default-on (fast-forward, no force-push). Live state now: mergeable: true (was dirty) — the conflict is cleared.

What the conflicts were: master's 2026-11-08 docs reorg moved root docs into docs/ (leaving redirect stubs), colliding with this PR's pre-move doc edits. Resolutions, per union of intent:

  • src/rollout.ts (the actual percent: 0 → 100 flip) + all three test files: merged cleanly, untouched.
  • Root CONFIGURATION.md / .zh-CN.md / PLUGIN.md: took master's redirect stubs (content now lives in docs/).
  • Ported the PR's "GA default-100%" wording from the old PLUGIN.md into docs/plugin-protocol.md (en + zh) so the relocated doc doesn't contradict the shipped default.
  • website/config-reference/rollout.yaml seed: kept master's watermark row + appended the pluginIncrementalTurns entry, then regenerated docs/configuration.md(+zh) / main.js / index.html via gen-config-docs (satisfies the docs-sync gate).
  • CHANGELOG.md: unioned master's new entries with the PR's #3248 entry.

Effective PR diff vs master = exactly the PR's original 9 files, no kernel/ (kernel-guard clean).

Honest caveats: (1) Local unit tests couldn't run here — this bare worktree has no node_modules/kernel/dist (environmental, not from the resolution); remote CI is authoritative. (2) Master advanced 37 commits during this run, so the PR now reports mergeable_state=behind (non-blocking — it still merges cleanly). I did not chase the moving master to avoid a re-stale/fight-for-the-branch loop; a final refresh pass at merge time is trivial if desired. Merge itself stays a human decision.


结构化八问

  1. 是什么问题:PR [PR] feat(config): ship the incremental turn lane at 100% — a fresh install is delta-by-default (#3248) #5486(增量车道 GA 默认 100%)与 master 冲突(merge state=dirty),无法合并。
  2. 什么原因导致的:master 于 2026-11-08 做文档重组——根目录文档迁入 docs/ 并留重定向占位;本 PR 的文档改动早于该重组、落在旧路径,两者在同一区域相撞。核心代码 src/rollout.ts(percent 0→100)与三个测试文件本身无冲突。
  3. 影响范围与严重程度:仅影响本 PR 能否合并,不影响线上行为;不解则一直 dirty 卡住合并。属合并阻塞,非运行缺陷、非缓存泄露。
  4. 怎么修复的:把当前 master 合入 PR 分支(merge、非 rebase、不强推),按 union-of-intent 逐文件解冲突:根目录三件取 master 占位;把「GA 默认 100%」措辞移植进 docs/plugin-protocol.md(中英);种子 yaml 保留 master 水位行并追加 pluginIncrementalTurns 条目后重新生成配置文档;CHANGELOG 取两侧并集。
  5. 是否有回归问题:功能 diff 未变(percent 0→100 + 测试),master 侧本就是绿的;逐项过高危模式——无语义漂移、无 fail-fast 变吞错、阈值/时序/输出格式未改(文档仅如实反映默认翻转为开)。
  6. 是否破坏缓存:不涉及(纯默认开关 + 文档,未触碰缓存/重试/游标逻辑)。
  7. 是否改了设计方向:不涉及(沿用既有分层门控,任一环节不满足即整轮回退全量重发)。
  8. 是否引入了新的配置:无新增配置面;BILI_ROLLOUT_PLUGIN_INCREMENTAL_TURNS 是 PR 既有逃生口,本次仅按规范登记进种子/生成文档。

判定:冲突已解并已推送(cb97026881),PR 现为 mergeable(原 dirty);运行中 master 又前进 37 提交故暂显 behind(不阻塞)。本地缺依赖未能跑单测(环境问题,以远程 CI 为准)。收工待 CI 复核 + 人工拍板合并。

Resolve website/config-reference/rollout.yaml conflict: master added desc_ja (Japanese) to the watermark entry; union keeps it and retains the PR's pluginIncrementalTurns entry, adding a desc_ja to that entry too (generator now validates desc_ja per entry and emits docs/configuration.ja.md; CI docs-sync checks it). Regenerated config docs.
@ranxianglei

Copy link
Copy Markdown
Owner

🤖 Powered by ework · qwen3.8-27b

[bot] CI fix — 2nd pass: re-merged master, resolved the re-conflict; PR is mergeable again

This run re-ran the conflict fix because master advanced again and re-dirtied the PR (3rd master SHA in ~33h).

What I did this run (2nd pass):

  • Fetched the fork head (cb97026881, my 1st-pass merge) + the new master head (fcc2e848); merged master into the PR branch.
  • Only one file conflicted: website/config-reference/rollout.yaml (the doc files auto-merged clean this time).
  • Root cause of the conflict: master added a desc_ja (Japanese) line to the pluginFullResendWatermark seed entry at the exact spot this PR appends its pluginIncrementalTurns entry.
  • Key finding: master now adds desc_ja to all seed entries, and tools/gen-config-docs.mjs validates that every entry has a desc_ja (it also emits a new docs/configuration.ja.md, which the CI docs-sync gate now checks). So the PR's new entry must carry a desc_ja or the generator/CI fails.
  • Resolution (union-of-intent): kept master's watermark desc_ja + kept the PR's pluginIncrementalTurns entry + added a Japanese desc_ja to that entry (matching its en/zh wording). Re-ran node tools/gen-config-docs.mjs generate (clean: 341 entries, i18n ok).
  • Committed the merge (ee0ed5d9f7) and pushed to fork branch 2026-11-01_incremental-lane-default-on (fast-forward, no force).

Live state (verified this run): PR #5486 → mergeable: true, mergeable_state: behind. The dirty conflict is resolved; "behind" only means master moved again after my push (fcc2e848 → 99c22d28) — it is not a conflict and not a merge blocker. Effective PR delta = 10 files (the original 9 + the new docs/configuration.ja.md), and it touches no kernel/ (kernel-guard stays clean).

Honesty note: local full-suite was not run (bare worktree, no node_modules); the merge is verified clean by the generator passing + clean auto-merge + no kernel/ in the delta. Full regression is the remote CI lane's job.

Note: master is moving very fast, so the PR will keep showing "behind" until it is merged; a final re-merge at merge time may be needed, but there are no conflicts left to resolve.

八问

  1. 是什么问题:PR 与 master 的合并冲突(merge_state=dirty)——master 快速推进后再次把 PR 打回冲突态。
  2. 什么原因导致的:master 给 rollout.yaml 的 pluginFullResendWatermark 种子条目加了 desc_ja(日文本地化),恰好落在这个 PR 追加 pluginIncrementalTurns 条目的同一插入点;且生成器现在强制校验每个条目必须有 desc_ja。
  3. 影响范围与严重程度:仅影响本 PR 能否合并(不阻塞其他 PR);不修则 PR 一直 dirty 无法合并。属"特定条件下有问题"(仅当 master 再推进时复发)。
  4. 怎么修复的:union-of-intent——保留 master 的 watermark desc_ja + 保留本 PR 的 pluginIncrementalTurns 条目,并给该条目补一条日文 desc_ja(与其 en/zh 措辞一致),再跑 gen-config-docs 重新生成文档。
  5. 是否有回归问题:无。纯合并 + 补本地化字段;src/rollout.ts、测试、docs/plugin-protocol.md 均干净自动合并;有效 diff 不含 kernel/(kernel-guard 不受影响)。
  6. 是否破坏缓存:不涉及(无缓存/重试/游标/去重键改动)。
  7. 是否改了设计方向:不涉及(无架构/边界/数据流转向)。
  8. 是否引入了新的配置:本次修复不引入新配置键——pluginIncrementalTurns 是本 PR 原有内容(默认 0→100 为 PR 既定意图、已审),本次只是给该条目补了生成器要求的 desc_ja 本地化字段。

判定:冲突已解决,PR 现为 mergeable(behind 非冲突、非合并阻塞)。收工。

@ranxianglei

Copy link
Copy Markdown
Owner

🤖 Powered by ework · qwen3.8-27b

[bot] 🏷 CI fix (2nd pass) — conflict resolved, PR mergeable. Outcome + 八问

(Confirming my immediately-preceding report with the required prefix — same run, same result.)

Done this run: re-merged master fcc2e848 into the PR branch (fork head cb97026881). One conflict — website/config-reference/rollout.yaml: master added a desc_ja (Japanese) line to the pluginFullResendWatermark seed entry at the PR's insertion point, and tools/gen-config-docs.mjs now validates desc_ja on every entry (also emitting docs/configuration.ja.md, now checked by the CI docs-sync gate). Resolved as union-of-intent + added a desc_ja to the PR's pluginIncrementalTurns entry; regenerated docs (clean, 341 entries). Pushed merge ee0ed5d9f7 to louie:2026-11-01_incremental-lane-default-on (fast-forward, no force).

Verified live this run: PR #5486 mergeable: true, mergeable_state: behind (master moved again to 99c22d28 — not a conflict, not a merge blocker). Delta = 10 files, no kernel/ (kernel-guard clean). Local full-suite not run (bare worktree, no node_modules); the merge is verified clean by the generator passing + clean auto-merge.

八问

  1. 是什么问题:PR 与 master 的合并冲突(dirty)——master 快速推进后再次把 PR 打回冲突态。
  2. 什么原因导致的:master 给 rollout.yaml 的 watermark 种子条目加了 desc_ja,恰落在本 PR 追加 pluginIncrementalTurns 条目的同一插入点;且生成器现在强制校验每个条目必须有 desc_ja。
  3. 影响范围与严重程度:仅影响本 PR 能否合并;不修则一直 dirty 无法合并。属"特定条件下有问题"(master 再推进才复发)。
  4. 怎么修复的:union-of-intent——保留 master 的 watermark desc_ja + 保留本 PR 条目 + 给该条目补一条 desc_ja,再跑 gen-config-docs 重新生成文档。
  5. 是否有回归问题:无。纯合并 + 补本地化字段;src/rollout.ts/测试/docs/plugin-protocol.md 干净自动合并;diff 不含 kernel/(kernel-guard 不受影响)。
  6. 是否破坏缓存:不涉及(无缓存/重试/游标/去重键改动)。
  7. 是否改了设计方向:不涉及(无架构/边界/数据流转向)。
  8. 是否引入了新的配置:本次修复不引入新键——pluginIncrementalTurns 是 PR 原有内容(0→100 为 PR 既定意图、已审),本次只补了生成器要求的 desc_ja 字段。

判定:冲突已解决,PR 现为 mergeable(behind 非冲突、非合并阻塞)。收工。

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

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants