Skip to content

[PR] fix: honest compose surfaces — [not foldable] tags on UNCOVERED HISTORY runs + historical-refs note on compress success receipts - #5311

Open
ranxianglei wants to merge 3 commits into
masterfrom
2026-11-01_compose-surface-honesty
Open

ranxianglei wants to merge 3 commits into
masterfrom
2026-11-01_compose-surface-honesty

Conversation

@ranxianglei

Copy link
Copy Markdown
Owner

Part of #5186 — compose-surface honesty (host side, proxy lane).

What

Two small host-side fixes for the "model composes a bad batch" family, both at the surfaces where ranges get composed:

  1. UNCOVERED HISTORY rows are tagged [not foldable] (src/uncovered.ts). The section prints live-looking ref spans (m02042–m02068 (27 msgs, ~7.1K tok)) with the "not as fold suggestions" caveat only in a trailing parenthetical — models copied the spans into compress batches anyway (independent codex repro in the CI queue starved for 8 hours by zombie runs: queued/running timeouts never cancelled, runs from deleted branches hold sl | CI 队列被僵尸 run 饿死 8 小时:排队/运行超时不取消、删分支的 run 永远占位 #5186 thread: a batch composed out of this exact line, whole batch rejected, 0 blocks folded). Every listed run now leads with [not foldable], matching the diagnostic-view marker family (⚠ not foldable: …). Both surfaces inherit it through the shared formatter: the acp_status section and the success-receipt "Uncovered history remains" verdict. Runs are by construction outside the actionable offer, so the tag is true for every row.

  2. Success receipts teach the source-of-truth split (src/stream.ts). Stored summaries quote refs verbatim and those quotes stay in context forever while the ledger can reset underneath them (session switch / native compaction restarts refs at m00001). Every fingerprinted compress success receipt now carries one line after the [Bug][pi] compress 回执指纹行被模型误读为摘要截断,触发防御性 decompress 浪费 token #2544 steering: "Refs quoted in stored summaries are historical snapshots: … Compose ranges only from acp_status output or the live ranges listed in this receipt." Rides the model-facing receipt AND the log copy (static text, no excerpts — 日志清理隐私信息 #1718 unaffected). Failure/no-op receipts are untouched.

Behavior changes (disclosed)

Tests

  • New tests/issue5186-compose-surface-honesty.test.ts: P1 formatter unit (tag on every listed run, cap intact), P2 acp_status surface renders the tag next to the hole span, N1 offered runs stay untagged (tag never leaks into the actionable list), P3 success receipt carries the note, N2 total-failure receipts don't.
  • tests/compress-fingerprint-refold.test.ts: receipt layout pins updated for the new line (one-note-per-line contract preserved); no-op receipt pinned to carry neither note.
  • Local targeted green: issue5186 (5/5), compress-fingerprint-refold (15/15), issue4331-uncovered-history (10/10), issue5139-turn-separator-ranges (4/4), issue847-compress-failure-guards (4/4); typecheck clean. Full-suite verdict from CI on this head.

@ranxianglei ranxianglei changed the title fix: honest compose surfaces — [not foldable] tags on UNCOVERED HISTORY runs + historical-refs note on compress success receipts [PR] fix: honest compose surfaces — [not foldable] tags on UNCOVERED HISTORY runs + historical-refs note on compress success receipts Oct 31, 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。 这是 #5186 compose-surface honesty 的 host 侧子件(代理车道),对应 GitHub PR #5311(分支 2026-11-01_compose-surface-honesty,单 commit,+178/−4,4 文件)。检索 open+closed 无并行 PR 触碰 src/uncovered.ts / 成功回执面;近邻条目均不同机制:已合入的 #4340(#4331 UNCOVERED HISTORY 段的来源)、#5279/#4410(插件车道 judge)、#4045(codex forge handoff)。

标签: bug + scope/wiring + p2 —— 缺陷类(模型把不可折叠行抄进 compress 批次导致整批被拒);改动面在请求管道/回执与 acp_status 工具表面,属协议接线;严重度 p2(有绕行:模型仍可只从 acp_status 组批,属浪费回合而非数据丢失/缓存泄露)。

正在对 PR #5311 做独立审核(方向、回归、代码、测试有效性),结论按结构化八问跟进。

@ranxianglei ranxianglei added bug Something isn't working p2 scope/wiring labels Oct 31, 2026
@github-actions

github-actions Bot commented Oct 31, 2026 •

Copy link
Copy Markdown
Contributor

📦 Built Package Artifact

Branch: 2026-11-01_compose-surface-honesty (138fa54)

Latest build — install this exact version

npm install -g billion-context@0.0.10984-pr.5311.8774576

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
138fa54 billion-context@0.0.10984-pr.5311.8774576 build
77ff8de billion-context@0.0.10896-pr.5311.bc17f7e build
ea67419 billion-context@0.0.10676-pr.5311.e9bc2fb build
82439e1 billion-context@0.0.9837-pr.5311.ed04290 build
12b4912 billion-context@0.0.9101-pr.5311.29902f8 build
21a6bec billion-context@0.0.9078-pr.5311.d2d7fba build
62e2595 billion-context@0.0.8326-pr.5311.e31ed17 build
205213e billion-context@0.0.8325-pr.5311.82baa43 build
140c4c5 billion-context@0.0.7437-pr.5311.433bfe8 build

Download artifact

  1. Download the artifact from the Actions run
  2. Extract the tarball and install:
tar xzf billion-context-pr5311.tgz
npm install -g package

Install instructions are refreshed on each push.

@ranxianglei

Copy link
Copy Markdown
Owner Author

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

The verdict-backed tag follow-up for the UNCOVERED HISTORY runs (the blanket [not foldable] tag at src/uncovered.ts:193) needs the apply-side projection entries (projectUnit / buildProjectionShared / collectCoverage / computeProtectedRefs) outside the kernel package boundary — they are not exported from kernel/src/index.ts today, and adding them is a kernel-boundary change that requires owner authorization before implementation.

Filed as #5784 (kernel authorization request: pure re-export, zero behavior change; options A approve / B keep the blanket tag). Verified against master @ be611a8: all four functions + the three types exist and are module-level exported, the package boundary genuinely lacks them, and flagFor's vocabulary in kernel/src/report.ts:453-476 is reusable verbatim for the per-run tags.

Waiting on the owner's A/B ruling; the verdict-tag follow-up lands on this branch after approval.

ranxianglei pushed a commit that referenced this pull request Nov 6, 2026
…surfaces

The compress success receipt gained the historical-refs note (#5311), which
enters the wire once per fold as an ephemeral tool payload and shifts the
session's input-token accounting by its length. The pinned outbound-body
digest of the fold turn, the persisted-state digests, and the stats/metadata
(+ nudge watermark) section digests move accordingly. Regenerated via
scripts/update-storage-render-goldens.ts under the documented parity pins;
A/B drive against the pre-change code shows exactly one differing wire body
(the fold turn carrying the note) and token-count-only state deltas.
@ranxianglei

Copy link
Copy Markdown
Owner Author

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

Addendum to my earlier note here: the verdict-backed screen no longer needs the #5784 kernel export — PR #5785 implements it host-side (core.applyCompression on a structuredClone of the state, zero diff under kernel/) and documents the coordination with this branch in its body (same parenthesized render hunk; whichever merges second carries a small rebase; the marker composes after any prefix).

The owner delegated the #5784 ruling and it is shelved (option B) as an optional optimization for any future surface that wants the direct-projection path. Nothing on this branch waits on #5784.

@ranxianglei

Copy link
Copy Markdown
Owner Author

🤖 Powered by ework · qwen3.8-27b

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

红 CI 已修复并全绿:新增 commit 62e2595b1(test: regenerate storage-render goldens)推到 PR 分支,新 head 的 30 项 check 全部通过(11 个 test 分片全 success,skip 项与之前一致)。

结构化八问

  1. 是什么问题:PR 给 compress 成功回执加了两处新文案(historical-refs 说明 + [not foldable] 标签),导致两个 golden 钉死套件(tests/golden-storage-render.test.ts、tests/storage-engine-proofs.test.ts)在 6 个测试分片上变红——它们钉的出站字节/持久化状态摘要没跟着更新。
  2. 什么原因导致的:golden 文件逐轮钉出站 body 摘要 + 持久化 state 摘要。新文案在 fold 那一轮以 ephemeral 工具载荷进一次线(该轮 body +292B),并把会话 input-token 账目平移了等长量 → fold 轮 body 摘要、state/stats/metadata(anthropic/responses 还含 nudge 水位键)各段摘要整体漂移。master(be611a8)同期全绿(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 今日多次确认),纯属本 PR 的 golden 滞后,非继承性失败。
  3. 影响范围与严重程度:仅 CI 红(30 项里 6 项),无运行时行为影响;阻塞合并。
  4. 怎么修复的:先做内容级 A/B 验证再动 golden——把两个生产文件临时还原到 master 版本、同一 drive 各跑一遍:15 轮出站 body 只有 fold 轮(release v0.1.2 #9)不同,差异恰好是那句 historical-refs note 原文,state 差异仅 token 计数(221331→221404);随后用官方脚本 scripts/update-storage-render-goldens.ts 按文档化 parity 参数重生成 4 份 golden,且重生成的 chat stateDigest(87edd26aa6797b9a)与 CI 失败输出里的 actual 值逐字一致——CI 环境等价性直接对上。单一 test: commit,diff 只含 4 个 golden JSON 的摘要字段,无任何结构变化。
  5. 是否有回归问题:无。diff 面 = 4 个测试夹具文件的哈希值;本地用 CI 同款 preload 参数复跑两个套件全绿(5/5、12/12);远程 CI 全绿收口。既有行为零触碰(未改任何生产代码、配置、kernel)。
  6. 是否破坏缓存:不涉及(golden 为测试夹具,与运行时缓存/重试/去重无关)。
  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。

备注:owner 评论里提到的 verdict-backed 标签后续项(需 kernel 边界导出,已立 #5784 待 A/B 拍板)不在本次红 CI 范围内,本次未触碰;golden 重生成不依赖该决策,blanket 标签版本即可先行合入。

判定:可合并。

@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)

Review verdict: APPROVE ✅ — clean host-side fix, correctly scoped, well-tested, green CI (30 checks, 0 failed/pending), zero kernel/ changes.
(Posted as a comment-review: GitHub forbids the author-identity from setting a formal APPROVE state on its own PR; the recommendation below stands and merge stays human.)

Reviewed the full diff (9 files) in source context at head 62e2595b1, both compression modes, and the golden-render regression surface.

Correctness (verified in source, not just the diff):

  • [not foldable] rides the shared describeUncoveredRuns; its only two callers are src/acp-status.ts:329 (UNCOVERED HISTORY section) and src/stream.ts:694 (success-receipt "Uncovered history remains" variant) → both surfaces tagged, exactly as claimed. Runs come from uncoveredHistoryRuns, which skips anything inside the actionable window (uncovered.ts:166), so the tag is true for every listed row.
  • HISTORICAL_REFS_NOTE is gated on changedBlocks.length > 0 (success-only) and rides both msg + logMsg (stream.ts:1595-1598) — consistent with #1718 (static text, no excerpts → log-safe; same pattern as the subagent note directly above it). Failure/no-op receipts untouched.

Regression audit:

  • Both modes: proxy-lane only. The proxy executes compress server-side (applyRanges); plugin-mode hosts execute locally and render their own receipts (KDD #7), so they get neither change. The tag is additive cosmetic text regardless.
  • Behavior drift old→new disclosed for both changes ✓.
  • Storage-render goldens: only the fold-round body (i:9) grew +292B (= the note); i:10/11/12 are byte-identical → the note appears once at the fold and does NOT propagate into the re-sent prefix → no recurring prefix-cache break. blockContents digest unchanged (stored summaries intact); state/stats/metadata/nudge digests moved as the expected consequence of the larger receipt. No wire-shape or persistence-format change.

Tests: real teeth — P1/P2/N1/P3/N2 positive+negative pairs (N1 pins the tag does NOT leak into the offered/actionable list; N2 pins failure receipts stay clean); compress-fingerprint-refold updated to pin one-note-per-line and that no-op carries neither note. These catch the regressions (tag removed / leaked into offers / note on failure).

Non-blocking note (transparency only): the PR body describes one new test file, but the diff also adds tests/issue5186-batch-coverage-staleness.test.ts (+171) pinning kernel applyCompression batch-vs-sequential coverage (tied to "#5186 checklist item ④ / #5780 body ④"; documents a probed non-reproduction + a nested-double-block wart). It is test-only, zero production-code change, within the #5186 family, and harmless — but it isn't mentioned in the body. Worth a one-line mention; not blocking.

Coordination (per owner addendum): noted — #5784 kernel export is moot (host-side via #5785, shelved option B); nothing on this branch waits on it. The shared render hunk with #5785 is a merge-ordering matter (second-merger rebases; marker composes after any prefix) and does not affect this PR's own diff.


八问(结构化结论)

  1. 是什么问题:模型会把「UNCOVERED HISTORY」里活像可折叠的 ref 段、以及旧摘要里引用的过期 ref,抄进 compress batch → 整批被拒、0 折叠。
  2. 什么原因导致的:两个诊断面(acp_status 未覆盖段 + 成功回执)把不可折叠的 span 用活 ref 呈现,且没告诉模型「摘要里引用的 ref 是随代际重置的历史快照」。
  3. 影响范围与严重程度:仅 proxy 车道;正确性/体验瑕疵(模型偶尔组错 batch),非崩溃、非数据丢失。
  4. 怎么修复的:在两个面上加诚实标注——每个未覆盖 run 前缀 [not foldable](共用 formatter,两处都生效);成功回执追加一行「引用是历史快照,range 只从 acp_status/本回执取」。改的是表面措辞,不是过滤模型输出。
  5. 是否有回归问题:无。仅折叠轮出站体 +292B(=该注记),后续轮字节不变 → 无重复前缀缓存破坏;blockContents 摘要字节不变;无 wire 形状/格式变更。
  6. 是否破坏缓存:不破坏前缀缓存(重发前缀不变);持久化字节随回执略增(已披露、有界、非格式变更)。
  7. 是否改了设计方向:否。仍是同源诊断面诚实化,无架构/边界转向。
  8. 是否引入了新的配置:否。无新 env/config/默认值。

合并判定:可合(APPROVE)。 唯一建议:PR 描述补一句第二个测试文件(非阻塞)。

@ranxianglei

Copy link
Copy Markdown
Owner Author

🤖 Powered by ework · qwen3.8-27b

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

Reviewed end-to-end; verdict: APPROVE. Posted the full review on the PR as a comment-review — GitHub won't let the author identity set a formal APPROVE state on its own PR, so the recommendation lives in the review body and merge stays yours.

Your addendum is incorporated and changes nothing blocking:

Regression checks I ran beyond CI-green (30 checks, 0 failed/pending):

  • Both surfaces confirmed through the shared formatter: src/acp-status.ts:329 (UNCOVERED HISTORY section) + src/stream.ts:694 (success-receipt uncovered variant). Runs come from uncoveredHistoryRuns, which skips anything inside the actionable window (uncovered.ts:166), so the tag is true for every listed row.
  • HISTORICAL_REFS_NOTE is success-only (changedBlocks.length > 0) and rides msg+logMsg consistently with 日志清理隐私信息 #1718 (static text, no excerpts → log-safe; same pattern as the subagent note directly above). Failure/no-op receipts untouched.
  • Storage-render goldens: only the fold-round body (i:9) grew +292B (= the note); i:10/11/12 are byte-identical → the note appears once at the fold and does NOT propagate into the re-sent prefix → no recurring prefix-cache break. blockContents digest unchanged (stored summaries intact); state/stats/metadata/nudge digests moved as the expected consequence of the larger receipt. No wire-shape or persistence-format change.

One non-blocking transparency nit: the PR body documents one new test file, but the diff also adds tests/issue5186-batch-coverage-staleness.test.ts (+171, test-only kernel applyCompression batch-vs-sequential pin, tied to #5186 item ④ / #5780 ④). It's harmless and in-family — just worth a one-line mention in the description. Not blocking.

Full 八问 structured conclusion is in the review body. 合并判定:可合(APPROVE)。

@ranxianglei
ranxianglei force-pushed the 2026-11-01_compose-surface-honesty branch from 62e2595 to 21a6bec Compare November 11, 2026 05:39
ranxianglei pushed a commit that referenced this pull request Nov 11, 2026
…surfaces

The compress success receipt gained the historical-refs note (#5311), which
enters the wire once per fold as an ephemeral tool payload and shifts the
session's input-token accounting by its length. The pinned outbound-body
digest of the fold turn, the persisted-state digests, and the stats/metadata
(+ nudge watermark) section digests move accordingly. Regenerated via
scripts/update-storage-render-goldens.ts under the documented parity pins;
A/B drive against the pre-change code shows exactly one differing wire body
(the fold turn carrying the note) and token-count-only state deltas.

Rebased onto master 33104b1 (conflict in these same four files: master had
regenerated them for its own changes). Post-rebase delta vs the pre-rebase
goldens is the metadata section digest only: master bumped the package
version 0.1.212 -> 0.1.213, which the persisted session metadata embeds.
Every per-turn wire body sha/shaCiteNorm is unchanged.
@ranxianglei
ranxianglei force-pushed the 2026-11-01_compose-surface-honesty branch from 21a6bec to 12b4912 Compare November 11, 2026 08:22
ranxianglei pushed a commit that referenced this pull request Nov 11, 2026
…surfaces

The compress success receipt gained the historical-refs note (#5311), which
enters the wire once per fold as an ephemeral tool payload and shifts the
session's input-token accounting by its length. The pinned outbound-body
digest of the fold turn, the persisted-state digests, and the stats/metadata
(+ nudge watermark) section digests move accordingly. Regenerated via
scripts/update-storage-render-goldens.ts under the documented parity pins;
A/B drive against the pre-change code shows exactly one differing wire body
(the fold turn carrying the note) and token-count-only state deltas.

Rebased onto master 33104b1 (conflict in these same four files: master had
regenerated them for its own changes). Post-rebase delta vs the pre-rebase
goldens is the metadata section digest only: master bumped the package
version 0.1.212 -> 0.1.213, which the persisted session metadata embeds.
Every per-turn wire body sha/shaCiteNorm is unchanged.

@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)

Re-review at new head 12b49120f — Verdict: APPROVE.
(Posted as a comment-review: GitHub blocks the author identity from setting a formal APPROVE on its own PR; the recommendation stands and merge stays human.)

What changed since my last review (62e2595b1): only a master merge into the branch (kernel → 0.0.195, #5765 blockOps gate, #6132 compressAtFoldPoint removal, …). The PR's own contribution vs master is byte-identical — same 9 files, same line deltas, same body, no kernel/ files. This is exactly the "second-merger rebases" coordination noted earlier.

Verified at the new head (re-checked, not assumed):

  • describeUncoveredRuns hunk byte-identical; still exactly two callers (src/acp-status.ts:339, src/stream.ts:711) — master added no third caller, so [not foldable] reaches precisely the two intended surfaces.
  • HISTORICAL_REFS_NOTE still gated on changedBlocks.length > 0 (success-only), riding msg+logMsg — consistent with #1718 (static text, no excerpts → log-safe). Failure/no-op receipts untouched.
  • The merged master introduced no semantic interaction with either surface (no new formatter caller; applyRanges gate intact).

Regression posture (unchanged from last round): proxy-lane only; plugin-mode hosts render their own receipts; the note appears once at the fold round (golden delta is the note-sized change at the single fold index; later bodies untouched) → no recurring prefix-cache break; stored-summary bytes intact; no wire-shape or persistence-format change. Behavior drift old→new disclosed in the body.

Tests: P1/P2/N1/P3/N2 positive+negative pairs pin the tag-on-every-run / tag-not-in-offer / note-on-success / note-not-on-failure directions; fingerprint-refold pins one-note-per-line and that no-op carries neither. These catch the regressions.

CI: at review time the new head shows 0 failures — 8 passed / 12 skipped / 10 guard-lanes still queued (suite in progress). The PR's own diff was fully green at the prior head and this merges onto green master.

No open findings. 合并判定:可合(APPROVE)。


八问(结构化结论)

  1. 是什么问题:模型会把「UNCOVERED HISTORY」里活像可折叠的 ref 段、以及旧摘要里引用的过期 ref,抄进 compress batch → 整批被拒、0 折叠。
  2. 什么原因导致的:两个诊断面(acp_status 未覆盖段 + 成功回执)把不可折叠 span 用活 ref 呈现,且未告知模型「摘要里引用的 ref 是随代际重置的历史快照」。
  3. 影响范围与严重程度:仅 proxy 车道;正确性/体验瑕疵(模型偶尔组错 batch),非崩溃、非数据丢失。
  4. 怎么修复的:在两个面上加诚实标注——每个未覆盖 run 前缀 [not foldable](共用 formatter,两处生效);成功回执追加一行「引用是历史快照,range 只从 acp_status/本回执取」。改的是表面措辞,不是过滤模型输出。
  5. 是否有回归问题:无。master 合并后逐点复核:formatter 仍仅两处调用、note 仍受成功门控、无新调用方/语义交互;仅折叠轮出站体增该注记、后续轮字节不变 → 无重复前缀缓存破坏;摘要字节不变;无 wire/格式变更。
  6. 是否破坏缓存:不破坏前缀缓存(重发前缀不变);持久化字节随回执略增(已披露、有界、非格式变更)。
  7. 是否改了设计方向:否。仍是同源诊断面诚实化,无架构/边界转向。
  8. 是否引入了新的配置:否。无新 env/config/默认值。

@ranxianglei

Copy link
Copy Markdown
Owner Author

🤖 Powered by ework · qwen3.8-27b

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

Re-reviewed at the new head 12b49120f. Verdict: APPROVE (posted as a comment-review — GitHub still blocks a formal self-APPROVE on the author's own PR; merge stays yours).

The only change since my last pass is the master merge you coordinated (kernel → 0.0.195, #5765 blockOps gate, #6132 compressAtFoldPoint removal). The PR's own 9-file diff vs master is byte-identical to what I reviewed — same hunks, same body, no kernel/. Re-verified at this head rather than assumed:

  • describeUncoveredRuns still has exactly two callers (src/acp-status.ts:339, src/stream.ts:711) — master added no third, so [not foldable] hits precisely the two intended surfaces.
  • HISTORICAL_REFS_NOTE still success-gated (changedBlocks.length > 0) on msg+logMsg; no semantic interaction from the merged master.

Regression posture unchanged: proxy-lane only; the note lands once at the fold round (later bodies untouched) → no recurring prefix-cache break; stored-summary / wire / persistence-format all intact. No open findings.

CI on this head: 0 failures at review time (8 passed / 12 skipped / 10 guard-lanes still queued — suite in progress); the PR's own diff was fully green at the prior head and merges onto green master.

Full 八问 structured conclusion is in the review. 合并判定:可合(APPROVE)。

@ranxianglei
ranxianglei force-pushed the 2026-11-01_compose-surface-honesty branch from 12b4912 to 82439e1 Compare November 14, 2026 03:07
ranxianglei pushed a commit that referenced this pull request Nov 14, 2026
…surfaces

The compress success receipt gained the historical-refs note (#5311), which
enters the wire once per fold as an ephemeral tool payload and shifts the
session's input-token accounting by its length. The pinned outbound-body
digest of the fold turn, the persisted-state digests, and the stats/metadata
(+ nudge watermark) section digests move accordingly. Regenerated via
scripts/update-storage-render-goldens.ts under the documented parity pins;
A/B drive against the pre-change code shows exactly one differing wire body
(the fold turn carrying the note) and token-count-only state deltas.

Rebased onto master 33104b1 (conflict in these same four files: master had
regenerated them for its own changes). Post-rebase delta vs the pre-rebase
goldens is the metadata section digest only: master bumped the package
version 0.1.212 -> 0.1.213, which the persisted session metadata embeds.
Every per-turn wire body sha/shaCiteNorm is unchanged.
@ranxianglei
ranxianglei force-pushed the 2026-11-01_compose-surface-honesty branch from 82439e1 to 14acdda Compare November 14, 2026 03:20
ranxianglei pushed a commit that referenced this pull request Nov 14, 2026
…surfaces

The compress success receipt gained the historical-refs note (#5311), which
enters the wire once per fold as an ephemeral tool payload and shifts the
session's input-token accounting by its length. The pinned outbound-body
digest of the fold turn, the persisted-state digests, and the stats/metadata
(+ nudge watermark) section digests move accordingly. Regenerated via
scripts/update-storage-render-goldens.ts under the documented parity pins;
A/B drive against the pre-change code shows exactly one differing wire body
(the fold turn carrying the note) and token-count-only state deltas.

Rebased onto master 33104b1 (conflict in these same four files: master had
regenerated them for its own changes). Post-rebase delta vs the pre-rebase
goldens is the metadata section digest only: master bumped the package
version 0.1.212 -> 0.1.213, which the persisted session metadata embeds.
Every per-turn wire body sha/shaCiteNorm is unchanged.

@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)

Re-review at head 14acddaa4 — Verdict: APPROVE (posted as a comment-review; a formal self-APPROVE is blocked because the posting PAT identity is the PR author ranxianglei).

What changed since my last review (head 12b49120f): another master re-merge into the branch — kernel 0.0.197, #6480 pure-SQLite default, #6640 watermark fastpath (#5403/#5418), #6508 announcement policy, #6568 release-gate, and more. The PR's own 3 commits (fix + 2 tests) were replayed on the new base.

The PR's contribution is unchanged (third confirmation). PR-vs-master (GitHub merge-base, base e5e1647ba0) is the same 9 files with the same line deltas, no kernel/:

  • src/stream.ts +7, src/uncovered.ts +6/-1, tests/compress-fingerprint-refold.test.ts +8/-3, 4 tests/golden/storage-render/*.json, 2 new test files (issue5186-compose-surface-honesty +157, issue5186-batch-coverage-staleness +171).

Re-verified at the new head (byte-identical hunks):

  • describeUncoveredRuns still has exactly two callers (src/acp-status.ts:339, src/stream.ts:711); the [not foldable] tag stays true for every listed run (they are outside the actionable offer by construction).
  • HISTORICAL_REFS_NOTE still success-gated — if (changedBlocks.length > 0) at src/stream.ts:1670, appended to msg + logMsg (1672/1673). Failure/no-op receipts untouched.
  • The 4 goldens still differ from the new master base by only the note-sized delta at the single fold index — master's storage/kernel changes (pure-SQLite default, kernel 0.0.197) are in the base, not the PR, so the note is the only PR-side persisted-state change. No recurring prefix-cache break (later round bodies byte-identical).

Regression posture unchanged: no wire/persistence-format change; proxy-lane only (plugin hosts render their own receipts); the note is static, excerpt-free, log-safe (#1718-consistent). The re-merge introduced no new regression.

⚠ CI state (honest note): at review time, head 14acddaa4 shows 25 cancelled / 3 success / 2 skipped / 0 failed / 0 pending in the check cache — the run looks cancelled/interrupted, not a clean green pass (the trigger header says "green", but the authoritative check source does not show one). The code verdict is on merit (byte-identical to a diff that was fully green at 62e2595b1); please confirm a green CI run on this head before merging.

No open code findings (NO OPEN ENDS): nothing to fix on the spot, nothing to convert to a decision, nothing parked.


8问 (Chinese, for human reviewers):

  1. 是什么问题: 让压缩回执对「不可折叠」与「历史引用」说真话——给 UNCOVERED HISTORY 每段加 [not foldable] 前缀,并在压缩成功回执里加一段「摘要里引用的 ref 是历史快照」的说明。
  2. 什么原因导致的: 模型把 UNCOVERED HISTORY 的 ref 区间误当折叠建议复制进 compress 范围;且会话重置(ref 从 m00001 重启)后,旧摘要里的 ref 引用可能失效,回执未提示。
  3. 影响范围与严重程度: 仅代理车道(proxy-lane)的压缩成功回执与 acp_status 的 UNCOVERED HISTORY 展示;属体验/诚实性瑕疵,无数据丢失、无 wire 变更。
  4. 怎么修复的: 在 describeUncoveredRuns 每行加 [not foldable] 前缀;新增 HISTORICAL_REFS_NOTE 常量,仅在压缩成功(changedBlocks.length > 0)时拼进 msg+logMsg。
  5. 是否有回归问题: 无。describeUncoveredRuns 恰好两个调用方,[not foldable] 对每行恒真;note 仅成功路径、静态无摘录、日志安全(#1718 一致);goldens 仅折叠轮 +292B,后续轮字节一致 → 无周期性前缀缓存破坏;rebase 后 master 侧存储/kernel 变更都在 base,PR 自身 diff 三次确认字节一致。
  6. 是否破坏缓存: 无。note 只落在单个折叠轮,后续轮字节不变。
  7. 是否改了设计方向: 无。不改架构/边界/数据流。
  8. 是否引入了新的配置: 无。

判定:APPROVE(代码层面);合并前请确认该 head 的 CI 为绿色(当前缓存显示 25 cancelled / 0 failed / 0 pending)。

@ranxianglei

Copy link
Copy Markdown
Owner Author

🤖 Powered by ework · qwen3.8-27b

[bot] 🏷 PR #5311 re-review (round 3) at head 14acddaa4 — Verdict: APPROVE (code merit).

This is the third review of this PR. Since my last review (12b49120f), the only change is another master re-merge (kernel 0.0.197, #6480 pure-SQLite default, #6640 watermark fastpath #5403/#5418, #6508 announcement policy, #6568 release-gate, …) with the PR's own 3 commits replayed on the new base.

The contribution is unchanged — third byte-identical confirmation. PR-vs-master (merge-base e5e1647ba0) = same 9 files, same deltas, no kernel/. Re-verified at the new head: describeUncoveredRuns still exactly two callers (src/acp-status.ts:339, src/stream.ts:711); HISTORICAL_REFS_NOTE still success-gated (if (changedBlocks.length > 0) at src/stream.ts:1670) on msg+logMsg; the 4 goldens differ from the new base by only the note-sized delta at the single fold index (master's storage/kernel changes are in the base, not the PR). No new regression.

⚠ CI is NOT green at this head. The check cache shows 25 cancelled / 3 success / 2 skipped / 0 failed / 0 pending (confirmed stable on a second read) — a cancelled/interrupted run, not the "green" the trigger header claims. The code verdict stands on merit (byte-identical to a diff that was fully green at 62e2595b1), but please confirm a green CI run on this head before merging.

Full review posted on the PR (review id 5901736025). No kernel/ → no 🔶 需要决策. No open code findings (NO OPEN ENDS).

8问 (Chinese):

  1. 是什么问题: 让压缩回执对「不可折叠」与「历史引用」说真话——给 UNCOVERED HISTORY 每段加 [not foldable] 前缀,并在压缩成功回执里加「摘要里引用的 ref 是历史快照」说明。
  2. 什么原因导致的: 模型把 UNCOVERED HISTORY 的 ref 区间误当折叠建议复制进 compress 范围;会话重置(ref 从 m00001 重启)后旧摘要 ref 可能失效,回执未提示。
  3. 影响范围与严重程度: 仅代理车道(proxy-lane)压缩成功回执与 acp_status 的 UNCOVERED HISTORY 展示;体验/诚实性瑕疵,无数据丢失、无 wire 变更。
  4. 怎么修复的: describeUncoveredRuns 每行加 [not foldable] 前缀;新增 HISTORICAL_REFS_NOTE,仅压缩成功(changedBlocks.length > 0)时拼进 msg+logMsg。
  5. 是否有回归问题: 无。describeUncoveredRuns 恰好两个调用方,[not foldable] 每行恒真;note 仅成功路径、静态无摘录、日志安全(日志清理隐私信息 #1718 一致);goldens 仅折叠轮 +292B,后续轮字节一致 → 无周期性前缀缓存破坏;PR 自身 diff 三次确认字节一致。
  6. 是否破坏缓存: 无。note 只落在单个折叠轮,后续轮字节不变。
  7. 是否改了设计方向: 无。
  8. 是否引入了新的配置: 无。

判定:APPROVE(代码层面);合并前请确认该 head 的 CI 为绿色(当前缓存 25 cancelled / 0 failed / 0 pending)。

@ranxianglei
ranxianglei force-pushed the 2026-11-01_compose-surface-honesty branch from 14acdda to ea67419 Compare November 17, 2026 20:21
ranxianglei pushed a commit that referenced this pull request Nov 17, 2026
…surfaces

The compress success receipt gained the historical-refs note (#5311), which
enters the wire once per fold as an ephemeral tool payload and shifts the
session's input-token accounting by its length. The pinned outbound-body
digest of the fold turn, the persisted-state digests, and the stats/metadata
(+ nudge watermark) section digests move accordingly. Regenerated via
scripts/update-storage-render-goldens.ts under the documented parity pins;
A/B drive against the pre-change code shows exactly one differing wire body
(the fold turn carrying the note) and token-count-only state deltas.

Rebased onto master 33104b1 (conflict in these same four files: master had
regenerated them for its own changes). Post-rebase delta vs the pre-rebase
goldens is the metadata section digest only: master bumped the package
version 0.1.212 -> 0.1.213, which the persisted session metadata embeds.
Every per-turn wire body sha/shaCiteNorm is unchanged.

@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)

Re-review at head ea674196e — Verdict: APPROVE (posted as a comment-review; a formal self-APPROVE is blocked because the posting PAT identity is the PR author ranxianglei).

What changed since my last review (head 14acddaa4): another master re-merge into the branch — #6996 compress-nudge-example, #7054 attribution-diagnostics topic, #6938 release-v0.1.216, #6958 Claude Code deferred-tools, #6922 architecture-docs map fix, #6953 reply-stash, wire-contract, and more (304 files). The PR's own 3 commits (fix + 2 tests) were replayed on the new base.

The PR's contribution is unchanged (fourth confirmation). PR-vs-master (GitHub merge-base, base a259e0b8fa) is the same 9 files with the same line deltas, no kernel/:

  • src/stream.ts +7, src/uncovered.ts +6/-1, tests/compress-fingerprint-refold.test.ts +8/-3, 4 tests/golden/storage-render/*.json, 2 new test files (issue5186-compose-surface-honesty +157, issue5186-batch-coverage-staleness +171).

Re-verified at the new head (byte-identical hunks; only line numbers shifted by master inserts):

  • describeUncoveredRuns still has exactly two callers (src/acp-status.ts:339, src/stream.ts:760); the [not foldable] tag stays true for every listed run (they are outside the actionable offer by construction).
  • HISTORICAL_REFS_NOTE still success-gated — if (changedBlocks.length > 0) at src/stream.ts:1791, appended to msg + logMsg (1793/1794). Failure/no-op receipts untouched.
  • The 4 goldens still differ from the new master base by only the note-sized delta at the single fold index (master's storage/kernel/wire changes are in the base, not the PR). No recurring prefix-cache break (later round bodies byte-identical).

Regression posture unchanged: no wire/persistence-format change; proxy-lane only (plugin hosts render their own receipts); the note is static, excerpt-free, log-safe (#1718-consistent). The re-merge introduced no new regression.

CI state (improved vs last round): head ea674196e is in progress — 14 pending (5 queued + 9 in_progress), 4 success, 12 skipped, 0 failed. The round-3 cancelled state is resolved; this is a live run with no failures so far. It should land green on completion; please confirm the final green before merging.

No open code findings (NO OPEN ENDS): nothing to fix on the spot, nothing to convert to a decision, nothing parked.


8问 (Chinese, for human reviewers):

  1. 是什么问题: 让压缩回执对「不可折叠」与「历史引用」说真话——给 UNCOVERED HISTORY 每段加 [not foldable] 前缀,并在压缩成功回执里加「摘要里引用的 ref 是历史快照」说明。
  2. 什么原因导致的: 模型把 UNCOVERED HISTORY 的 ref 区间误当折叠建议复制进 compress 范围;会话重置(ref 从 m00001 重启)后旧摘要 ref 可能失效,回执未提示。
  3. 影响范围与严重程度: 仅代理车道(proxy-lane)压缩成功回执与 acp_status 的 UNCOVERED HISTORY 展示;体验/诚实性瑕疵,无数据丢失、无 wire 变更。
  4. 怎么修复的: describeUncoveredRuns 每行加 [not foldable] 前缀;新增 HISTORICAL_REFS_NOTE,仅压缩成功(changedBlocks.length > 0)时拼进 msg+logMsg。
  5. 是否有回归问题: 无。describeUncoveredRuns 恰好两个调用方,[not foldable] 每行恒真;note 仅成功路径、静态无摘录、日志安全(#1718 一致);goldens 仅折叠轮 +292B,后续轮字节一致 → 无周期性前缀缓存破坏;PR 自身 diff 四次确认字节一致。
  6. 是否破坏缓存: 无。note 只落在单个折叠轮,后续轮字节不变。
  7. 是否改了设计方向: 无。
  8. 是否引入了新的配置: 无。

判定:APPROVE(代码层面);CI 进行中(0 failed),完成后确认绿色即可合并。

@ranxianglei

Copy link
Copy Markdown
Owner Author

🤖 Powered by ework · qwen3.8-27b

[bot] 🏷 PR #5311 re-review (round 4) at head ea674196e — Verdict: APPROVE (code merit).

Fourth review of this PR. Since my last review (14acddaa4), the only change is another master re-merge (#6996 compress-nudge-example, #7054 attribution-diagnostics, #6938 release-v0.1.216, #6958 CC deferred-tools, #6922 architecture-docs, #6953 reply-stash, wire-contract, … — 304 files) with the PR's own 3 commits replayed on the new base.

The contribution is unchanged — fourth byte-identical confirmation. PR-vs-master (merge-base a259e0b8fa) = same 9 files, same deltas, no kernel/. Re-verified at the new head (only line numbers shifted by master inserts): describeUncoveredRuns still exactly two callers (src/acp-status.ts:339, src/stream.ts:760); HISTORICAL_REFS_NOTE still success-gated (if (changedBlocks.length > 0) at src/stream.ts:1791) on msg+logMsg; the 4 goldens differ from the new base by only the note-sized delta at the single fold index. No new regression.

CI is healthy this round (improved vs last round's cancelled state): head ea674196e is in progress — 14 pending (5 queued + 9 in_progress), 4 success, 12 skipped, 0 failed. A live run with no failures so far; should land green on completion. Please confirm the final green before merging.

Full review posted on the PR (review id 5946257807). No kernel/ → no 🔶 需要决策. No open code findings (NO OPEN ENDS).

8问 (Chinese):

  1. 是什么问题: 让压缩回执对「不可折叠」与「历史引用」说真话——给 UNCOVERED HISTORY 每段加 [not foldable] 前缀,并在压缩成功回执里加「摘要里引用的 ref 是历史快照」说明。
  2. 什么原因导致的: 模型把 UNCOVERED HISTORY 的 ref 区间误当折叠建议复制进 compress 范围;会话重置(ref 从 m00001 重启)后旧摘要 ref 可能失效,回执未提示。
  3. 影响范围与严重程度: 仅代理车道(proxy-lane)压缩成功回执与 acp_status 的 UNCOVERED HISTORY 展示;体验/诚实性瑕疵,无数据丢失、无 wire 变更。
  4. 怎么修复的: describeUncoveredRuns 每行加 [not foldable] 前缀;新增 HISTORICAL_REFS_NOTE,仅压缩成功(changedBlocks.length > 0)时拼进 msg+logMsg。
  5. 是否有回归问题: 无。describeUncoveredRuns 恰好两个调用方,[not foldable] 每行恒真;note 仅成功路径、静态无摘录、日志安全(日志清理隐私信息 #1718 一致);goldens 仅折叠轮 +292B,后续轮字节一致 → 无周期性前缀缓存破坏;PR 自身 diff 四次确认字节一致。
  6. 是否破坏缓存: 无。note 只落在单个折叠轮,后续轮字节不变。
  7. 是否改了设计方向: 无。
  8. 是否引入了新的配置: 无。

判定:APPROVE(代码层面);CI 进行中(0 failed),完成后确认绿色即可合并。

@ranxianglei
ranxianglei force-pushed the 2026-11-01_compose-surface-honesty branch from ea67419 to 77ff8de Compare November 18, 2026 21:35
ranxianglei pushed a commit that referenced this pull request Nov 18, 2026
…surfaces

The compress success receipt gained the historical-refs note (#5311), which
enters the wire once per fold as an ephemeral tool payload and shifts the
session's input-token accounting by its length. The pinned outbound-body
digest of the fold turn, the persisted-state digests, and the stats/metadata
(+ nudge watermark) section digests move accordingly. Regenerated via
scripts/update-storage-render-goldens.ts under the documented parity pins;
A/B drive against the pre-change code shows exactly one differing wire body
(the fold turn carrying the note) and token-count-only state deltas.

Rebased onto master 33104b1 (conflict in these same four files: master had
regenerated them for its own changes). Post-rebase delta vs the pre-rebase
goldens is the metadata section digest only: master bumped the package
version 0.1.212 -> 0.1.213, which the persisted session metadata embeds.
Every per-turn wire body sha/shaCiteNorm is unchanged.
@ranxianglei

Copy link
Copy Markdown
Owner Author

🤖 Powered by ework · qwen3.8-27b

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

CI 工程师本轮结论:上报的冲突已解决,本轮无需再 rebase / push。

触发器报「PR 与 base 冲突 (merge state: dirty)」,但该状态来自 11-17 的过期镜像缓存(缓存 sha 仍是旧 head ea674196e)。实查 live 状态:该分支在本轮会话开始前几分钟(05:30–05:32 +0800)已被一次并行/前序 run rebase 到当前 master 并推送。我未再改动、未再次 push(远端已处于正确终态)。

本轮核实的证据(全部 live,非记忆):

  • origin/master = a446a3c1c;PR 分支 2026-11-01_compose-surface-honesty = 77ff8deaf。
  • merge-base = a446a3c1c = master head;分支 3 ahead / 0 behind → master 是 PR 分支的祖先,干净 fast-forward,无冲突(git merge-tree --write-tree 退出 0,无冲突消息)。
  • 3 个 commit 时间戳 2026-11-19 05:30:58 / 05:30:58 / 05:32:41 +0800 —— 本轮(05:38 起)之前刚推。
  • PR-vs-master diff = 预期的 9 个文件(src/stream.ts、src/uncovered.ts、tests/compress-fingerprint-refold.test.ts、4 份 storage-render golden、tests/issue5186-batch-coverage-staleness.test.ts、tests/issue5186-compose-surface-honesty.test.ts),与前几轮字节同族,无 kernel/。
  • 终 head 77ff8deaf 的 commit message 不含 [ework:ci-held] → 这是最终 head,CI 应已触发。

CI 状态说明(诚实边界): 镜像缓存仍停在旧 head ea674196e(fetched 11-17,17 success / 13 skipped / 0 failed),尚未刷新到新 head 77ff8deaf,因此新 head 的 CI 绿/红我无法从缓存核实——镜像轮询器(~60s)追上后即可见。旧 head 全绿,且新 head 是同样 3 个 commit 在新 master 上的 rebase,预期等价;合并前请以新 head 的最终绿色 run 为准。

结构化八问

  1. 是什么问题:上游 PR [PR] fix: honest compose surfaces — [not foldable] tags on UNCOVERED HISTORY runs + historical-refs note on compress success receipts #5311 被报「与 base 分支冲突 (merge state: dirty)」,阻塞合并。
  2. 什么原因导致的:master 前进、PR 分支落后产生冲突;但该冲突已在本轮会话开始前几分钟(05:30–05:32)被一次并行/前序 run rebase 解决并推送,触发器依据的是 11-17 的过期缓存。
  3. 影响范围与严重程度:仅 PR 可合并性(冲突阻塞合并);无运行时行为影响。
  4. 怎么修复的:未重复 rebase——核实分支已被 rebase 到当前 master 并推送;验证 merge-base=master head、3 ahead/0 behind、git merge-tree 干净 fast-forward 无冲突。
  5. 是否有回归问题:无。PR diff 为预期 9 文件(与前几轮同族),无 kernel/,合并为干净 fast-forward。
  6. 是否破坏缓存:不涉及(纯 rebase/冲突解决,未触碰缓存机制)。
  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。

判定:冲突已解决(由前序/并行 run 完成),分支干净且已 rebase 到当前 master,本轮无遗留 rebase/push 动作;新 head CI 绿/红待镜像缓存刷新后确认。

@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)

Re-review at head 77ff8deaf — Verdict: APPROVE (posted as a comment-review; a formal self-APPROVE is blocked because the posting PAT identity is the PR author ranxianglei).

What changed since my last review (head ea674196e): another master re-merge into the branch. The PR's own 3 commits (fix + 2 tests) were replayed on the new base.

The PR's contribution is unchanged (fifth confirmation). PR-vs-master (GitHub merge-base, base a446a3c1c8) is the same 9 files with the same line deltas, no kernel/:

  • src/stream.ts +7, src/uncovered.ts +6/-1, tests/compress-fingerprint-refold.test.ts +8/-3, 4 tests/golden/storage-render/*.json, 2 new test files (issue5186-compose-surface-honesty +157, issue5186-batch-coverage-staleness +171).

Re-verified at the new head (byte-identical hunks; only line numbers shifted by master inserts):

  • describeUncoveredRuns still has exactly two callers (src/acp-status.ts:339, src/stream.ts:774); the [not foldable] tag stays true for every listed run (they are outside the actionable offer by construction).
  • HISTORICAL_REFS_NOTE still success-gated — if (changedBlocks.length > 0) at src/stream.ts:1815, appended to msg + logMsg (1817/1818). Failure/no-op receipts untouched.
  • The 4 goldens still differ from the new master base by only the note-sized delta at the single fold index (master's changes are in the base, not the PR). No recurring prefix-cache break (later round bodies byte-identical).

Regression posture unchanged: no wire/persistence-format change; proxy-lane only (plugin hosts render their own receipts); the note is static, excerpt-free, log-safe (#1718-consistent). The re-merge introduced no new regression.

CI is GREEN this round: head 77ff8deaf shows 17 success / 13 skipped / 0 failed / 0 pending — a clean green pass. This is the first fully-green CI state across my five reviews (round 3 was cancelled, round 4 in-progress). Ready to merge from a CI standpoint.

No open code findings (NO OPEN ENDS): nothing to fix on the spot, nothing to convert to a decision, nothing parked.


8问 (Chinese, for human reviewers):

  1. 是什么问题: 让压缩回执对「不可折叠」与「历史引用」说真话——给 UNCOVERED HISTORY 每段加 [not foldable] 前缀,并在压缩成功回执里加「摘要里引用的 ref 是历史快照」说明。
  2. 什么原因导致的: 模型把 UNCOVERED HISTORY 的 ref 区间误当折叠建议复制进 compress 范围;会话重置(ref 从 m00001 重启)后旧摘要 ref 可能失效,回执未提示。
  3. 影响范围与严重程度: 仅代理车道(proxy-lane)压缩成功回执与 acp_status 的 UNCOVERED HISTORY 展示;体验/诚实性瑕疵,无数据丢失、无 wire 变更。
  4. 怎么修复的: describeUncoveredRuns 每行加 [not foldable] 前缀;新增 HISTORICAL_REFS_NOTE,仅压缩成功(changedBlocks.length > 0)时拼进 msg+logMsg。
  5. 是否有回归问题: 无。describeUncoveredRuns 恰好两个调用方,[not foldable] 每行恒真;note 仅成功路径、静态无摘录、日志安全(#1718 一致);goldens 仅折叠轮 +292B,后续轮字节一致 → 无周期性前缀缓存破坏;PR 自身 diff 五次确认字节一致。
  6. 是否破坏缓存: 无。note 只落在单个折叠轮,后续轮字节不变。
  7. 是否改了设计方向: 无。
  8. 是否引入了新的配置: 无。

判定:APPROVE(代码层面);CI 已全绿(17 success / 13 skipped / 0 failed),可合并。

@ranxianglei

Copy link
Copy Markdown
Owner Author

🤖 Powered by ework · qwen3.8-27b

[bot] 🏷 PR #5311 re-review (round 5) at head 77ff8deaf — Verdict: APPROVE (code merit), CI green.

Fifth review of this PR. Since my last review (ea674196e), the only change is another master re-merge, with the PR's own 3 commits replayed on the new base.

The contribution is unchanged — fifth byte-identical confirmation. PR-vs-master (merge-base a446a3c1c8) = same 9 files, same deltas, no kernel/. Re-verified at the new head (only line numbers shifted by master inserts): describeUncoveredRuns still exactly two callers (src/acp-status.ts:339, src/stream.ts:774); HISTORICAL_REFS_NOTE still success-gated (if (changedBlocks.length > 0) at src/stream.ts:1815) on msg+logMsg; the 4 goldens differ from the new base by only the note-sized delta at the single fold index. No new regression.

CI is GREEN this round: head 77ff8deaf shows 17 success / 13 skipped / 0 failed / 0 pending — a clean green pass (first fully-green state across my five reviews; round 3 cancelled, round 4 in-progress). Ready to merge from a CI standpoint.

Full review posted on the PR (review id 5965795449). No kernel/ → no 🔶 需要决策. No open code findings (NO OPEN ENDS).

8问 (Chinese):

  1. 是什么问题: 让压缩回执对「不可折叠」与「历史引用」说真话——给 UNCOVERED HISTORY 每段加 [not foldable] 前缀,并在压缩成功回执里加「摘要里引用的 ref 是历史快照」说明。
  2. 什么原因导致的: 模型把 UNCOVERED HISTORY 的 ref 区间误当折叠建议复制进 compress 范围;会话重置(ref 从 m00001 重启)后旧摘要 ref 可能失效,回执未提示。
  3. 影响范围与严重程度: 仅代理车道(proxy-lane)压缩成功回执与 acp_status 的 UNCOVERED HISTORY 展示;体验/诚实性瑕疵,无数据丢失、无 wire 变更。
  4. 怎么修复的: describeUncoveredRuns 每行加 [not foldable] 前缀;新增 HISTORICAL_REFS_NOTE,仅压缩成功(changedBlocks.length > 0)时拼进 msg+logMsg。
  5. 是否有回归问题: 无。describeUncoveredRuns 恰好两个调用方,[not foldable] 每行恒真;note 仅成功路径、静态无摘录、日志安全(日志清理隐私信息 #1718 一致);goldens 仅折叠轮 +292B,后续轮字节一致 → 无周期性前缀缓存破坏;PR 自身 diff 五次确认字节一致。
  6. 是否破坏缓存: 无。note 只落在单个折叠轮,后续轮字节不变。
  7. 是否改了设计方向: 无。
  8. 是否引入了新的配置: 无。

判定:APPROVE(代码层面);CI 已全绿(17 success / 13 skipped / 0 failed),可合并。

ci-check added 3 commits November 19, 2026 20:03
…d refs as historical on compress success receipts (#5186)
…: no observable staleness divergence; one nested-double-block wart pinned) (#5186)
…surfaces

The compress success receipt gained the historical-refs note (#5311), which
enters the wire once per fold as an ephemeral tool payload and shifts the
session's input-token accounting by its length. The pinned outbound-body
digest of the fold turn, the persisted-state digests, and the stats/metadata
(+ nudge watermark) section digests move accordingly. Regenerated via
scripts/update-storage-render-goldens.ts under the documented parity pins;
A/B drive against the pre-change code shows exactly one differing wire body
(the fold turn carrying the note) and token-count-only state deltas.

Rebased onto master 33104b1 (conflict in these same four files: master had
regenerated them for its own changes). Post-rebase delta vs the pre-rebase
goldens is the metadata section digest only: master bumped the package
version 0.1.212 -> 0.1.213, which the persisted session metadata embeds.
Every per-turn wire body sha/shaCiteNorm is unchanged.
@ranxianglei
ranxianglei force-pushed the 2026-11-01_compose-surface-honesty branch from 77ff8de to 138fa54 Compare November 19, 2026 12:14
@ranxianglei

Copy link
Copy Markdown
Owner Author

🤖 Powered by ework · qwen3.8-27b

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

CI 工程师本轮:冲突已解决,rebase + golden 重生后 force-push 新 head 138fa54d4。

本轮的冲突是真实的(与上轮不同):master 前进了 21 个 commit,其中 f357310ed(test(golden): regenerate storage-render goldens — systemFp/canonicalSysFp now persist the 256B change-probe head, #7149 batch2)也重生了同 4 份 storage-render golden,与本 PR 的 golden 重生 commit(77ff8deaf)撞车 → 4 个内容级冲突(tests/golden/storage-render/{anthropic,chat,google,responses}.json)。

本轮做的(全部 live 核实):

  1. git fetch 确认 master 已移到 6453feeeb,PR 分支落后 21 / 领先 3;git merge-tree 确认 4 个 golden 冲突(仅 golden,src/stream.ts/src/uncovered.ts master 未触碰 → 前两个 commit 干净重放)。
  2. 从 PR head 拉工作分支 ci-5311-rb,git rebase origin/master:commit 1(fix)、commit 2(test pin)干净重放;commit 3(golden 重生)在 4 个 golden 上冲突。
  3. 冲突解法 = 官方脚本重生,非手工拼块:装依赖、build kernel + pi-subagents,跑 node --import tsx scripts/update-storage-render-goldens.ts(脚本自带 parity 参数 BILI_IMAGE_RESCALE=0/BILI_SNAPSHOT_SHADOW=0/BILI_MESSAGE_TIMESTAMPS=0)在 rebase 后的树上重生成 4 份 golden —— 同时吸收 master 的 change-probe 变更与本 PR 的 historical-refs note delta。
  4. 验证:重生后 golden 对 master 的 diff 恰好是本 PR 的 note 尺寸 delta(折叠轮 body +292B:49770→50062/49700→49992/52585→52877,state/stats/metadata/nudge 摘要随之平移),与前几轮字节同族;tests/golden-storage-render(5/5)+ storage-engine-proofs + issue5186(10/10)+ compress-fingerprint-refold + issue4331/issue5139/issue847 共 60/60 定向测试全绿;npm run typecheck 干净(exit 0)。
  5. rebase 完成:merge-base = master head,3 ahead / 0 behind,git merge-tree exit 0(干净 fast-forward,无冲突);PR-vs-master diff = 预期 9 文件,无 kernel/。
  6. push 前再 git fetch 确认远端 PR 分支仍停在 rebase 前 head 77ff8deaf(无并发推送)→ git push --force-with-lease 到新 head 138fa54d4(本任务明确授权的 conflict-rebase force-push;终 head 不带 [ework:ci-held],触发一次完整 CI)。

CI 状态(诚实边界): 镜像缓存仍停在 rebase 前的旧 head 77ff8deaf(fetched 11-18 22:01,17 success / 13 skipped / 0 failed,mergeable_state 因当时落后仍标 blocked),尚未刷新到新 head 138fa54d4——等新 head 的最终绿色 run 以镜像缓存为准。旧 head 全绿 + 新 head 是同样 3 commit 在新 master 上的 rebase + golden 已按官方脚本重生 + 本地 60/60 定向绿,预期等价转绿。

结构化八问

  1. 是什么问题:PR [PR] fix: honest compose surfaces — [not foldable] tags on UNCOVERED HISTORY runs + historical-refs note on compress success receipts #5311 与 base 分支冲突(merge state: dirty)——master 前进 21 commit,其中 f357310ed 重生同 4 份 storage-render golden,与本 PR 的 golden 重生撞车,阻塞合并。
  2. 什么原因导致的:双方各自重生了同一组 golden 文件(本 PR 为 CI queue starved for 8 hours by zombie runs: queued/running timeouts never cancelled, runs from deleted branches hold sl | CI 队列被僵尸 run 饿死 8 小时:排队/运行超时不取消、删分支的 run 永远占位 #5186 historical-refs note;master f357310ed 为 性能清理提案:同一份内容每请求被重复哈希 3-5 次 —— 盘点结论与分批清理方案 #7149 batch2 change-probe head 持久化),内容级冲突。
  3. 影响范围与严重程度:仅 PR 可合并性(4 个 golden 文件冲突阻塞合并);无运行时行为影响(golden 是测试夹具)。
  4. 怎么修复的:rebase 到当前 master(6453feeeb);4 个 golden 冲突用官方脚本 scripts/update-storage-render-goldens.ts 在 rebase 后的树上重生成——同时吸收 master 的 change-probe 变更与本 PR 的 note delta,非手工拼冲突块。
  5. 是否有回归问题:无。重生后 golden 对 master 的 diff 恰好是本 PR 的 note 尺寸 delta(折叠轮 +292B + 摘要平移),与前几轮字节同族;60/60 定向测试绿、typecheck 干净;无 kernel/ 变更。
  6. 是否破坏缓存:不涉及(golden 为测试夹具;note 只落在单个折叠轮,后续轮字节不变——前几轮已确认无周期性前缀缓存破坏)。
  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。

判定:冲突已解决(rebase + 官方脚本 golden 重生),分支干净 fast-forward,已 force-push 新 head 138fa54d4;新 head CI 绿/红待镜像缓存刷新后确认(当前缓存仍停在 rebase 前旧 head 77ff8deaf,该旧 head 全绿 17 success / 13 skipped / 0 failed)。

@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)

Re-review at head 138fa54d4 — Verdict: APPROVE (posted as a comment-review; a formal self-APPROVE is blocked because the posting PAT identity is the PR author ranxianglei).

What changed since my last review (head 77ff8deaf): another master re-merge into the branch. The PR's own 3 commits (fix + 2 tests) were replayed on the new base.

The PR's contribution is unchanged (sixth confirmation). PR-vs-master (GitHub merge-base, base 6453feeeb8) is the same 9 files with the same line deltas, no kernel/:

  • src/stream.ts +7, src/uncovered.ts +6/-1, tests/compress-fingerprint-refold.test.ts +8/-3, 4 tests/golden/storage-render/*.json, 2 new test files (issue5186-compose-surface-honesty +157, issue5186-batch-coverage-staleness +171).

Re-verified at the new head (byte-identical hunks; line numbers stable vs last round):

  • describeUncoveredRuns still has exactly two callers (src/acp-status.ts:339, src/stream.ts:774); the [not foldable] tag stays true for every listed run (they are outside the actionable offer by construction).
  • HISTORICAL_REFS_NOTE still success-gated — if (changedBlocks.length > 0) at src/stream.ts:1815, appended to msg + logMsg (1817/1818). Failure/no-op receipts untouched.
  • The 4 goldens still differ from the new master base by only the note-sized delta at the single fold index (master's changes are in the base, not the PR). No recurring prefix-cache break (later round bodies byte-identical).

Regression posture unchanged: no wire/persistence-format change; proxy-lane only (plugin hosts render their own receipts); the note is static, excerpt-free, log-safe (#1718-consistent). The re-merge introduced no new regression.

CI is GREEN: head 138fa54d4 shows 17 success / 13 skipped / 0 failed / 0 pending — a clean green pass. Ready to merge.

No open code findings (NO OPEN ENDS): nothing to fix on the spot, nothing to convert to a decision, nothing parked.


8问 (Chinese, for human reviewers):

  1. 是什么问题: 让压缩回执对「不可折叠」与「历史引用」说真话——给 UNCOVERED HISTORY 每段加 [not foldable] 前缀,并在压缩成功回执里加「摘要里引用的 ref 是历史快照」说明。
  2. 什么原因导致的: 模型把 UNCOVERED HISTORY 的 ref 区间误当折叠建议复制进 compress 范围;会话重置(ref 从 m00001 重启)后旧摘要 ref 可能失效,回执未提示。
  3. 影响范围与严重程度: 仅代理车道(proxy-lane)压缩成功回执与 acp_status 的 UNCOVERED HISTORY 展示;体验/诚实性瑕疵,无数据丢失、无 wire 变更。
  4. 怎么修复的: describeUncoveredRuns 每行加 [not foldable] 前缀;新增 HISTORICAL_REFS_NOTE,仅压缩成功(changedBlocks.length > 0)时拼进 msg+logMsg。
  5. 是否有回归问题: 无。describeUncoveredRuns 恰好两个调用方,[not foldable] 每行恒真;note 仅成功路径、静态无摘录、日志安全(#1718 一致);goldens 仅折叠轮 +292B,后续轮字节一致 → 无周期性前缀缓存破坏;PR 自身 diff 六次确认字节一致。
  6. 是否破坏缓存: 无。note 只落在单个折叠轮,后续轮字节不变。
  7. 是否改了设计方向: 无。
  8. 是否引入了新的配置: 无。

判定:APPROVE(代码层面);CI 已全绿(17 success / 13 skipped / 0 failed),可合并。

@ranxianglei

Copy link
Copy Markdown
Owner Author

🤖 Powered by ework · qwen3.8-27b

[bot] 🏷 PR #5311 re-review (round 6) at head 138fa54d4 — Verdict: APPROVE (code merit), CI green.

Sixth review of this PR. Since my last review (77ff8deaf), the only change is another master re-merge, with the PR's own 3 commits replayed on the new base.

The contribution is unchanged — sixth byte-identical confirmation. PR-vs-master (merge-base 6453feeeb8) = same 9 files, same deltas, no kernel/. Re-verified at the new head (line numbers stable vs last round): describeUncoveredRuns still exactly two callers (src/acp-status.ts:339, src/stream.ts:774); HISTORICAL_REFS_NOTE still success-gated (if (changedBlocks.length > 0) at src/stream.ts:1815) on msg+logMsg; the 4 goldens differ from the new base by only the note-sized delta at the single fold index. No new regression.

CI is GREEN: head 138fa54d4 shows 17 success / 13 skipped / 0 failed / 0 pending — a clean green pass. Ready to merge.

Full review posted on the PR (review id 5975321993). No kernel/ → no 🔶 需要决策. No open code findings (NO OPEN ENDS).

8问 (Chinese):

  1. 是什么问题: 让压缩回执对「不可折叠」与「历史引用」说真话——给 UNCOVERED HISTORY 每段加 [not foldable] 前缀,并在压缩成功回执里加「摘要里引用的 ref 是历史快照」说明。
  2. 什么原因导致的: 模型把 UNCOVERED HISTORY 的 ref 区间误当折叠建议复制进 compress 范围;会话重置(ref 从 m00001 重启)后旧摘要 ref 可能失效,回执未提示。
  3. 影响范围与严重程度: 仅代理车道(proxy-lane)压缩成功回执与 acp_status 的 UNCOVERED HISTORY 展示;体验/诚实性瑕疵,无数据丢失、无 wire 变更。
  4. 怎么修复的: describeUncoveredRuns 每行加 [not foldable] 前缀;新增 HISTORICAL_REFS_NOTE,仅压缩成功(changedBlocks.length > 0)时拼进 msg+logMsg。
  5. 是否有回归问题: 无。describeUncoveredRuns 恰好两个调用方,[not foldable] 每行恒真;note 仅成功路径、静态无摘录、日志安全(日志清理隐私信息 #1718 一致);goldens 仅折叠轮 +292B,后续轮字节一致 → 无周期性前缀缓存破坏;PR 自身 diff 六次确认字节一致。
  6. 是否破坏缓存: 无。note 只落在单个折叠轮,后续轮字节不变。
  7. 是否改了设计方向: 无。
  8. 是否引入了新的配置: 无。

判定:APPROVE(代码层面);CI 已全绿(17 success / 13 skipped / 0 failed),可合并。

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 p2 scope/wiring

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant