Skip to content

[PR] Index content's FK child columns — releaseContent DELETE stops full-scanning children - #5540

Merged
ranxianglei merged 9 commits into
masterfrom
2026-11-02_fk-content-indexes
Nov 5, 2026
Merged

ranxianglei merged 9 commits into
masterfrom
2026-11-02_fk-content-indexes

Conversation

@ranxianglei

Copy link
Copy Markdown
Owner

Closes #5539

What and why

With foreign_keys=ON, deleting a parent content row requires SQLite to prove no child row still references it — it does that by scanning each child table BY ITS REF COLUMN: an index lookup when indexed, a full scan when not. Six columns reference content(hash) — messages.content_hash, blocks.one_hash, blocks.summary_hash, ccr_entries.content_hash, message_media.media_hash, checkpoints.state_hash — and none carried an index, so every releaseContent DELETE paid six full scans. Busy sessions release ~3K hashes per turn (field report #5539: 20–35 s/turn at ~18K messages).

This adds all six indexes as schema v11, following the #4800 precedent (content(refcount), v9): idempotent creates in the existing migration transaction for existing libraries, DDL lines for fresh installs, plus the mysql-side equivalent.

Sixth index beyond the field audit: #5539 enumerates five child columns; checkpoints.state_hash is the same mechanism missed there (it is also scanned on every parent delete). Same failure domain, so it ships in this fix instead of a split.

Persistence surface (§7.4 flag)

Surface Change Compat
sqlite schema version v10 → v11 additive-only: six new indexes. Older builds ignore extra indexes; newer builds create them at open via the from < 11 step (idempotent CREATE INDEX IF NOT EXISTS).
mysql stamp v10 → v11 existing tables gain the keys via probe-first ALTER TABLE … ADD INDEX inside the existing current < MYSQL_SCHEMA_VERSION gate; stamp uses GREATEST(…) so it never downgrades. Fresh installs get the KEY lines from the DDL.
wire / message shape none —
content addressing / refcount semantics none FK enforcement itself is unchanged: deleting a still-referenced parent throws exactly as before (regression-tested both directions).

MySQL honesty note: the mysql DDL declares NO FOREIGN KEY constraints, so InnoDB performs no FK-presence checks on delete — mysql users do not exhibit this pathology today. The mysql keys are parity + future-proofing (if FKs are ever declared), costing nothing and keeping the two backends' index sets aligned.

Behavior change disclosure (old → new + why)

Behavior Old New Why
releaseContent per-parent-delete plan 6 full child-table scans 6 index searches the scans proved absence of references by brute force; indexes make the proof O(log n)
turn latency at scale (field, ~18K messages) 20–35 s/turn 4–5 s/turn (reporter's production measurement after adding the keys) root cause removed, not masked
open/migration cost — one-time build of six small indexes on upgrade (same order as the existing open-time quick_check gate) bounded one-off
db.prepare() calls in releaseContent 2 per released hash (~3K/turn) prepared once per store instance, reused field-measured prepare overhead ≈7× the statement execution (0.063 vs 0.009 ms/call); second commit, behavior identical

Local repro (node:sqlite, 20K content rows + six child tables, exact releaseContent shape, single transaction): BEFORE 0.210 ms/delete, plan = SCAN ×6; AFTER 0.012 ms/delete, plan = SEARCH USING INDEX ×6 (≈17×). FK enforcement verified unchanged both directions (referenced parent still throws; after the child is gone the delete succeeds).

Tests

  • NEW tests/issue5539-fk-content-indexes.test.ts (gate discipline — both directions): fresh install has v11 + all six indexes and the DELETE plan shows SEARCH USING <index> per child and no SCAN; a library stamped v10 WITHOUT the indexes (precondition asserts the plan actually scans) migrates to v11 with the improved plan on reopen; FK enforcement still rejects deleting a referenced parent.
  • Version-pin assertions updated to the new head in seven existing storage suites.
  • Targeted local runs green: issue5539 3/3 · issue4796-gc-batched-sweep 8/8 · storage-unified · storage-ext · storage-ingest-watermark · issue4666 · issue4677 · storage-mysql-seam; typecheck clean. Full-suite verdict = CI.

Coexistence

src/storage/store.ts is a hot file (#3987/#4242 lineage touches releaseContent); the second commit is a self-contained lazy-cache inside that one method and stands independently of the schema commits.

ci-check added 2 commits November 2, 2026 21:36
…er full-scans (#5539)

With foreign_keys=ON, deleting a parent content row must prove no child row still references it; SQLite proves that by scanning each child table BY ITS REF COLUMN — an index lookup when indexed, a full table scan when not. Six columns reference content(hash) (messages.content_hash, blocks.one_hash, blocks.summary_hash, ccr_entries.content_hash, message_media.media_hash, checkpoints.state_hash) and none carried an index, so every releaseContent DELETE paid six full scans. Busy sessions release ~3K hashes per turn — field report #5539: 20-35 s/turn at ~18K messages, 4-5 s after adding the keys.

Schema v11 (sqlite + mysql):
- STORAGE_DDL_V1 / MYSQL_DDL carry the six new indexes for fresh installs
- migrateSchema from<11 block: idempotent CREATE INDEX x6 (v9 discipline)
- mysql init(): STATISTICS probe widened to all six tables + probe-first ALTER ADD INDEX x6
- version-pin assertions in seven existing suites moved to the new head

Local repro (20K messages, one transaction, exact releaseContent shape):
0.210 ms/delete before (plan: SCAN x6) vs 0.012 ms/delete after (plan:
SEARCH USING INDEX x6); FK enforcement itself unchanged (a referenced
parent still throws FOREIGN KEY constraint failed).
The field report measured per-call db.prepare() at ~7x the cost of the
statement itself (0.063 vs 0.009 ms/call) — releaseContent runs once per
released hash (~3K/turn on busy sessions), so the prepare overhead is a
real share of the hot path even after the index fix removes the scans.

Statements are now prepared lazily once per UnifiedStore instance and
reused. Safe because each store owns exactly one connection for its whole
life (private readonly db, closed in close(), never re-opened on the same
instance). Behavior unchanged: same SQL, same ordering, same FK semantics.
@github-actions

github-actions Bot commented Nov 2, 2026 •

Copy link
Copy Markdown
Contributor

📦 Built Package Artifact

Branch: 2026-11-02_fk-content-indexes (f502174)

Latest build — install this exact version

npm install -g billion-context@0.0.8245-pr.5540.e77458a

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
f502174 billion-context@0.0.8245-pr.5540.e77458a build
d1cb503 billion-context@0.0.8242-pr.5540.40d1298 build
2d51f4d billion-context@0.0.7892-pr.5540.bf6d987 build
eae2078 billion-context@0.0.7885-pr.5540.502a728 build
a1e10fc billion-context@0.0.7883-pr.5540.5316abe build
b01fa03 billion-context@0.0.7882-pr.5540.01e5203 build

Download artifact

  1. Download the artifact from the Actions run
  2. Extract the tarball and install:
tar xzf billion-context-pr5540.tgz
npm install -g package

Install instructions are refreshed on each push.

@ranxianglei ranxianglei changed the title Index content's FK child columns — releaseContent DELETE stops full-scanning children [PR] Index content's FK child columns — releaseContent DELETE stops full-scanning children Nov 2, 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 #5539(字段报告:releaseContent 每回合 ~3K 次父行 DELETE × 6 张无索引子表全表扫描,~18K 消息会话 20–35 s/turn)的修复 PR。扫了全部 open PR:唯一触碰同一机制(content 的 FK 子列建索引 / schema v11)的就是本 PR;#5250(second-writer degrade)是写者争用域,机制不同。无并列在途 PR。

标签: bug scope/storage p2(性能缺陷、有触发条件——繁忙大会话;非缓存泄露类,不升 p1)。

现在做三件事,随后按八问体例给完整 review 结论:① 逐段读 diff(schema/store/mysql/测试)并核对既有行为;② 本地定向跑新回归测试 + 相关 storage 套件;③ 查 CI 状态。

@ranxianglei ranxianglei added bug Something isn't working p2 scope/storage labels Nov 2, 2026
ci-check added 3 commits November 2, 2026 21:55
…eep (#5539)

The four H.stamp() assertions in the BILI_MYSQL_TEST_DSN-gated subtests of
tests/issue4666-migration-dedup.test.ts still expected the pre-#5539 head
version '10'; they skip locally without a MySQL DSN, so only the remote
mysql-contract lane caught them ('11' !== '10'). Now expect '11'.
Review nit (#5542): the releaseContent statement-cache hunk accidentally
re-indented the untouched class field lastIntegrityOk from 4 to 9 spaces.
Revert to 4 so the diff carries no incidental whitespace changes.
Review nit (#5542): the v10 -> v11 bump updated asserts but left three
comments claiming v10 state that no longer holds — a fresh install now runs
the v11 DDL, and the idempotence arms reopen v11-stamped databases.
@ranxianglei

Copy link
Copy Markdown
Owner Author

🤖 Powered by ework · qwen3.8-27b

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

Review 结论:可合(等最终 head eae2078d0 的 CI 绿;涉及存储 schema 版本 bump,按 §7.4 属人工合并档)。

结构化八问

  1. 是什么问题:foreign_keys=ON 下每次删除 content 父行,SQLite 都要证明「没有子行引用它」;六个 FK 子列无索引 → 每删一行就对六张子表各做一次全表扫描。releaseContent(每轮约 3K 次父行删除)在大会话(~18K 消息)上单轮卡 20–35 秒([perf] FK child columns of content are unindexed — every releaseContent DELETE full-scans messages #5539 现场报告)。
  2. 什么原因导致的:REFERENCES content(hash) 的六个子列建表时就没配索引,SQLite FK 校验在无索引时只能线性扫子表。扫描成本随表大小线性增长,所以只在大会话上暴露——机制一直在,只是量级到了才疼。
  3. 影响范围与严重程度:仅统一存储表达到数万行的大会话;症状是轮次延迟,不是数据丢失、不是缓存泄露;小会话几乎无感。p2(已打标)。
  4. 怎么修复的:给六个 FK 子列补索引(v11 迁移 CREATE INDEX IF NOT EXISTS,fresh DDL 同步带上);MySQL 侧补 KEY 行 + 逐条守卫的 ALTER(MySQL 不声明 FK,属对等/未来防护,PR 正文已披露)。顺带把 releaseContent 的每次调用 prepare 改为实例级懒准备一次、复用同一语句(SQL 与调用顺序不变)。
  5. 是否有回归问题:无。逐项审过的既有行为:
    • 迁移路径(fresh / v10 库 / 更早库):受影响的全部测试套件本地绿(env-gated 的 arm 本地 skip,CI 覆盖);
    • FK 强制语义不变——索引只改「证明成本」不改约束行为,测试 3 双向钉住(被引用父行删除仍抛 FOREIGN KEY constraint failed、删完子行后成功、foreign_key_check 为空);
    • releaseContent 语义一致(同语句、同顺序、null-guard 保留);db 句柄生命周期核实过(构造时建立、close() 关闭一次),缓存语句不会活得比连接久;
    • MySQL probe-first 单次 STATISTICS 查询 + 每索引独立守卫 = 幂等、崩溃安全(stamp 最后落);
    • 版本 pin 清点:作者 follow-up a1e10fcd1 补齐 mysql-contract pin 后,我全仓 grep 复查,无残留功能性旧 pin。
  6. 是否破坏缓存:不涉及。纯存储层变更,出站字节不动,prefix cache 不受影响;语句缓存是进程内性能优化,不改任何线上行为。
  7. 是否改了设计方向:不涉及——沿 v9/v10 既有迁移纪律的加法式演进。
  8. 是否引入新配置:不涉及。无新 env/config key。

审中我直接推了两个小修到 PR 分支(独立标注 commit,小问题直接修):

  • 9c0a4d244:src/storage/store.ts:432 一处无关字段被顺带重排缩进(4→9 空格),还原(纯空白);
  • eae2078d0:版本 sweep 漏掉的三处注释事实("v10"→"v11",tests/storage-ext.test.ts:184、tests/storage-unified.test.ts:841/:866),改对。
    两 commit 为空白/注释级:typecheck exit 0,受影响套件本地全绿。

测试有效性:新回归套件带双向门(#3617 纪律)——先断言未加索引的 v10 计划确实全扫(负向对照),再断言 EXPLAIN QUERY PLAN 命中全部六个索引且无 SCAN <table>;FK arm 双向钉住。是真断言,回归时抓得住。

CI 状态:写作时对 pre-nit head a1e10fcd1 的 run 仍在跑;我的两个 commit 会触发新一轮。master 漂移 +3(kernel #5531 修复等)与本 PR 零文件重叠,可干净合并(GitHub 显示 behind,非阻塞)。

判定:可合,等 eae2078d0 CI 绿后由人工合并。

@ranxianglei

Copy link
Copy Markdown
Owner Author

🤖 Powered by ework · qwen3.8-27b

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

收工小结(完整八问 review 见上条评论):

  • 审核:机制核实(FK 子列无索引 → 每次父行删除全扫六张子表,v11 补六个索引 + releaseContent 语句缓存),迁移幂等性、FK 语义双向、版本 pin 清点全部过,无回归;新回归套件带负向对照,是真断言。
  • 审中小修已推 PR 分支:9c0a4d244(store.ts:432 无关缩进还原)、eae2078d0(三处 sweep 漏掉的 v10→v11 注释),空白/注释级,typecheck + 受影响套件本地绿。
  • CI(最终 head eae2078d0):全部 guard/contract 车道绿(static-gates、mysql-contract、kernel-guard、version-guard、lock-sync、docs-sync 等),12 个 test shard 仍在跑,目前 0 failed。
  • 标签:bug / scope/storage / p2(大会话轮次延迟,非数据丢失/缓存泄露)。

判定维持:可合,等 test shard 全绿后人工合并(schema 版本 bump,§7.4 人工档)。

…ap case (#5539)

The six FK-child indexes added by schema v11 append index-maintenance frames to every insert on messages/blocks/ccr_entries/message_media/checkpoints. In the #3250 custom-cap sequence the post-reset tail therefore grew from 54.3 KiB locally / 66.4 KiB on CI (pre-v11) to 130.8-146.9 KiB (1182472-1198952; identical 1198952 on ubuntu-24 shard 3), past the 128 KiB budget. Same drift class as #4506 and #4796/#5193 (documented in-test): every checkpoint reset still truncates exactly to the configured cap (verified per-ingest on both trees), so the inter-reset budget widens 128 -> 256 KiB with provenance; the bound stays far below the ~4.4 MiB default-limit plateau the assertion discriminates against.
…ual (#4599)

01fcc5d (#5656/#5584, merged via #5687) diverts tool-less dsh single-shot
requests onto the verbatim side lane (kernel no-touch); the residual for main
turn 1 is documented in the commit's own comment. Five pre-existing pins
assumed turn 1 rides the full pipeline and went born-red at merge (master red
since; every PR head inherits it).

Test-only realignment codifying the documented behavior:
- issue2419/2446/2447/2481 guard-stamp subtests: two turns; the BLOCKED ref
  is witnessed on the first FULL-PIPELINE turn (turn 2). No fold can happen
  before turn 2 (nothing compressible, no action surface), so protection
  against every possible fold is unchanged.
- issue2817: stats.requests off by one (turn 1 uncounted); the persona-fold
  teeth are intact, assertions shifted -1.
- issue5403 A/B golden: turn 0 rides the side lane in BOTH phases (the gate
  ignores the rollout env), so rb[0] is unstamped by design; stamps resume at
  turn 1. Byte equality holds at i=0 (both phases forward verbatim).

No src/, kernel/, config, or version changes. Locally green: all six files +
side-lane-invariants + issue5656-single-shot-sidecar + dsh-persona-fingerprint.
@ranxianglei

Copy link
Copy Markdown
Owner Author

🤖 Powered by ework · qwen3.8-27b

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

结论先行:本次红是继承性失败(master 自身红),收敛修复已在分支上(另一车道先于本车道完成合并);本轮工作 = 核实修复正确性 + 确认新 head 全绿,未重复推送。

核实链(证据):

  1. 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(CI 总控):master 红,根因 = 五枚 dsh sidecar 回归钉(DSH: If the first request of a session originates from a bypass call, compression for the entire session is permanently | DSH: If the first request of a session originates from a bypass call, compress for the entire session is permanently inv | DSH:会话首个… #5656 残留行为),已收敛到 ci-fixes@0470bec62(PR [PR] test(ci): converge master red — realign five dsh pins to the #5656 single-shot sidecar residual (#4599) #5718 待 owner 合并)。
  2. 红 run 日志(run 42012406213,ubuntu-24-shard-2,共 fail 2):① tests/issue2419-dsh-skill-catalog.test.ts(ci-fixes 修复的 6 个测试文件之一,报错行 the catalog message must carry a BLOCKED ref (never advertised, never foldable))= 继承签名;② tests/msgid-lane-flow.test.ts(已知负载 flake,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 floor 956/969 同签名)。两者均与本 PR 的 diff(FK 子列索引 / releaseContent DELETE)无关。
  3. 分支已在修复中前进:新 head f5021747a(11:13Z,ci-check 车道推送)= 红 head d1cb50384 + 仅 ci-fixes 单提交;git diff 逐行核对两 head 差异恰好是那 6 个 dsh 测试文件(86+/49−),无其他内容夹带。
  4. 新 head CI 全绿:主 CI run 42017499693 全部 11 个 test shard + 全部门禁 success;其余 10 条 workflow(e2e 各车道、build artifact 等)亦全部 success。

七问:

  1. 是什么问题:PR [PR] Index content's FK child columns — releaseContent DELETE stops full-scanning children #5540 的 11 个 test 分片中 10 个全红(head d1cb503),静态门禁全绿。
  2. 什么原因导致的:继承性失败——master 自身带红(五枚 dsh 回归钉与 DSH: If the first request of a session originates from a bypass call, compression for the entire session is permanently | DSH: If the first request of a session originates from a bypass call, compress for the entire session is permanently inv | DSH:会话首个… #5656 sidecar 残留行为不一致),本 PR 旧 head 基于收敛前的 master、不含 ci-fixes@0470bec62;与本 PR 代码改动无关(日志失败签名逐条对上继承家族 + 已知 flake)。
  3. 影响范围与严重程度:仅检查矩阵信号级(所有继承 master 红的 PR 同受影响);无用户可见功能影响;本 PR 自身的 FK 索引优化不受牵连。
  4. 怎么修复的:按 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 政策「merge origin/ci-fixes 进 PR 分支」收敛——该步已由另一车道(ci-check)先行完成(f502174);本车道核实了两 head 差异纯净(仅 6 测试文件)与新 head CI 全绿。再推一次只是重复空操作,故不推。
  5. 是否有回归问题:无。两 head 间 delta 仅为 test-only 提交(把五枚钉对齐到当前行为),零 src 改动;本 PR 自身 diff 未被触碰。
  6. 是否破坏缓存:不涉及。
  7. 是否引入新配置:不涉及(设计方向亦未变)。

判定:CI blocker 已清除,PR 就绪待 owner 合并。备注:镜像 check 缓存截至我读取时仍显示旧 head(fetched_at 10:50Z),若后续再有指向旧 sha 的「仍红」唤醒,按陈旧缓存处理即可。

@ranxianglei

Copy link
Copy Markdown
Owner Author

🤖 Powered by ework · qwen3.8-27b

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

Review verdict: APPROVE(可合) — human merge(§7.4:schema 版本 bump 属持久化格式层,不 auto-merge)。Head f5021747a,CI 全绿(镜像缓存 2026-11-05T11:48Z:30 checks、0 failed,mysql-contract success——正是验证 pin sweep 的 lane)。

独立审核说明:未照抄前轮结论——自己重读了全 diff(schema.ts / store.ts / mysql.ts / 新测试 / 全部 pin 更新)、枚举了 DDL 里所有 REFERENCES content(hash) 列、本地定向跑了受影响套件 + typecheck。

结构化八问:

  1. 是什么问题:releaseContent 每轮逐个删除 content 父行(~3K/turn);foreign_keys=ON 下 SQLite 删父行前必须证明没有子行还引用该 hash,而六个 FK 子列全无索引,每个父行删除触发 6 次子表全表扫描——现场实测 ~18K messages 时 20–35 s/turn。
  2. 什么原因导致的:机制级根因——SQLite FK 强制按引用列扫子表,无索引只能全扫;原 DDL 从未建这六个索引(现场审计列了五个,本修复补齐同机制的第六个 checkpoints.state_hash)。修的是根因(索引缺失),不是症状。
  3. 影响范围与严重程度:所有 unified-storage(SQLite)实例的忙碌长会话;严重度=每轮延迟劣化(20–35 s/turn),非数据丢失、非不可用。MySQL 后端不声明 FOREIGN KEY 约束,无此病理(其上的 key 是 parity + 未来防护,PR 已如实披露)。
  4. 怎么修复的:schema v10→v11 建六个幂等 CREATE INDEX IF NOT EXISTS(fresh DDL 全量 + 迁移事务块内,同一纪律如 v9);MySQL 侧镜像:fresh DDL 带 key + probe-first ALTER(单次 STATISTICS sweep,GREATEST stamp 防降级);次要发现落地:releaseContent 两条语句改为每 store 实例 prepare 一次(现场实测 prepare ≈7× exec)。
  5. 是否有回归问题:无。逐项清点既有行为:
  6. 是否破坏缓存:不涉及(存储层索引,不触碰 prefix cache/重试/增量游标;建索引不改任何行字节)。
  7. 是否改了设计方向:不涉及(纯增量 schema 演进 + 语句复用,无边界/数据流转向)。
  8. 是否引入新配置:不涉及(无新 env/config/默认值;STORAGE_SCHEMA_VERSION 10→11 是内部常量,非配置面)。

测试有效性:新测试有牙——前置条件先断言未索引计划确实 SCAN(模拟不复现 bug 就红),再断言迁移后 EXPLAIN QUERY PLAN 命中全部六个索引且无任何 SCAN <table>(双向,#3617 门纪律),第三个钉 FK 强制双向。本地定向验证:8 个存储相关套件(新测试、#3250、#4666、#4677、#4796、watermark、ext、unified)全绿 + typecheck clean;全量判定 = CI(绿)。

行为变更清单(披露):

未发现需要顺手修的小问题(storage-unified.test.ts:843 标题「schema v10」命名的是被测功能代际,与「schema v2」等 house style 一致,断言已正确指向 v11,保留)。

判定:APPROVE / 可合,人工合并(persistence-version 层)。

@ranxianglei

Copy link
Copy Markdown
Owner Author

🤖 Powered by ework · qwen3.8-27b

[bot] 🏷 (ai: vllm-qwen/qwen3.8-27b)

Review lane 收工确认:PR #5540(head f502174)已 end-to-end 独立审核完毕,完整八问审核结论见上方评论——verdict:APPROVE(可合),人工合并(§7.4 persistence-version 层)。要点:六个 FK 子列索引覆盖完整(枚举 DDL 验证)、v10→v11 纯增量迁移幂等、releaseContent SQL 逐字不变且语句缓存安全、新测试双向有牙、回归清点无破坏、行为变更已披露(6 全表扫描→6 索引查找;20–35s→4–5s/turn)。CI 镜像缓存全绿(30 checks,0 failed,mysql-contract success);本地定向套件 + typecheck 绿。无顺手修项,无 kernel/ 触碰。协调行已报 #4599。

@ranxianglei
ranxianglei merged commit bfabcd1 into master Nov 5, 2026
49 checks passed
@github-actions

github-actions Bot commented Nov 5, 2026

Copy link
Copy Markdown
Contributor

📦 Merged Build Published

This merge to master was published as 0.0.1044-master.bfabcd1 (rolling master tag).

npm install -g billion-context@0.0.1044-master.bfabcd1  # this exact build
npm install -g billion-context@master  # always the newest merged build

npm view billion-context@master prints the current version at any time.

ranxianglei pushed a commit to mariopasquali/billion-context that referenced this pull request Nov 5, 2026
…anxianglei#4542/etc (conflict resolution for ranxianglei#5743)

# Conflicts:
#	tests/opencode-rolling-cache-marks.test.ts
Louie0806 added a commit to Louie0806/billion-context that referenced this pull request Nov 5, 2026
…dexes and the v11 stamp sweep

master advanced 9a2d51d -> bfabd1ca with six commits from ranxianglei#5539/ranxianglei#5540:
the releaseContent FK child indexes, cached prepared statements, the v11
stamp re-sweep, and a WAL-tail budget for the ranxianglei#3250 custom-cap case.

One textual conflict, in tests/issue4666-migration-dedup.test.ts: this lane
pins the mysql stamp to STORAGE_SCHEMA_VERSION (added when ranxianglei#4796 left one pin
behind), while master re-swept the same lines to the literal "11". Kept the
derived form - it survives the next version bump, which is the whole reason
this lane introduced it - and refreshed the stale ranxianglei#4796 comment to mention
ranxianglei#5539 as well.
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/storage

Projects

None yet

Development

Successfully merging this pull request may close these issues.

[perf] FK child columns of content are unindexed — every releaseContent DELETE full-scans messages

1 participant