From 64dcf238efb764554a649bf69fcb9dd8f98fcad8 Mon Sep 17 00:00:00 2001 From: LoopX Agent <337587101+loopx-agent@users.noreply.github.com> Date: Thu, 8 Oct 2026 23:54:07 +0800 Subject: [PATCH] docs(authority): preserve formal SQLite capacity qualification gap Signed-off-by: LoopX Agent <337587101+loopx-agent@users.noreply.github.com> --- .../2026-09-28-retirement-cadence.md | 35 +++++++++++++++++-- .../2026-09-28-retirement-cadence.zh-CN.md | 29 +++++++++++++-- 2 files changed, 60 insertions(+), 4 deletions(-) diff --git a/docs/architecture/rfcs/ledger/shared-goal-authority-state-provider-v0/2026-09-28-retirement-cadence.md b/docs/architecture/rfcs/ledger/shared-goal-authority-state-provider-v0/2026-09-28-retirement-cadence.md index 0a79892fd7..32009604f1 100644 --- a/docs/architecture/rfcs/ledger/shared-goal-authority-state-provider-v0/2026-09-28-retirement-cadence.md +++ b/docs/architecture/rfcs/ledger/shared-goal-authority-state-provider-v0/2026-09-28-retirement-cadence.md @@ -458,8 +458,39 @@ fresh processes, crossing a checkpoint and checking original-receipt replay, changed-intent rejection and projection/hash parity at every step. It qualifies that bounded storage journey, **not** Host execution, live Goal adoption or D2's ten-day soak. No active authority, release default or legacy-writer deletion -decision changes. B still needs sustained workload/platform/capacity evidence; -C still needs consumer/onboarding and supported upgrade acceptance. +decision changes. C still needs consumer/onboarding and supported upgrade +acceptance. + +The unchanged `matched-64k` capacity runner at `5f51559dc` completed independent +10k/100k histories on macOS arm64, Node 24.21.0 / SQLite 3.53.4, with WAL/FULL +durability and the same non-empty 1,341-file public scan root. Its formal ledger +is **13 passed / 1 failed / 11 missing**; the process exits 1 and +`full_d2_qualified=false`. Selected p95 measurements are: + +| Measurement | 10k | 100k | Frozen budget / result | +| --- | ---: | ---: | --- | +| Warm head | 2.360 ms | 4.860 ms | Growth 2.059x > 2x: **failed**; absolute < 50 ms passes | +| Commit | 4.654 ms | 7.022 ms | Growth 1.509x and absolute < 100 ms pass | +| Receipt | 5.043 ms | 6.952 ms | Growth 1.379x and absolute < 50 ms pass | +| Fresh CLI status | 1,133.177 ms | 1,143.820 ms | < 2,000 ms passes | +| Fresh CLI mutation | 671.668 ms | 865.628 ms | Increment 193.960 ms < 200 ms passes | + +Real CLI operations, head/receipt checks and temporary-store cleanup completed; +no correctness assertion failed. These storage axes do not qualify the complete +eight-agent/four-writer workload, steady-state RSS, large-history recovery, +consumer lag, upgrade/rollback, ten-day soak, supported runtime matrix, promotion, +1 MiB payload, 300k headroom or 60-second burst. No OS cache flush was performed. +[D2 issue #4224](https://github.com/loopx-project/loopx/issues/4224) already reports +a soak start; completion and applicability to this candidate need verification +with its existing owner before authorizing any replacement run. + +The next bounded work isolates continuity-query, proof/codec, connection and host +costs before choosing a repair. The root cause is unproven. Keep the original +failed result and budgets; do not remove hole/digest/receipt verification, shorten +the workload or rerun until green. A passing absolute head budget does not cancel +its failed growth budget. B remains open; this single-source result neither +establishes a release regression nor changes provider defaults or File comparison +requirements. ### Contract health follows Todo authority diff --git a/docs/architecture/rfcs/ledger/shared-goal-authority-state-provider-v0/2026-09-28-retirement-cadence.zh-CN.md b/docs/architecture/rfcs/ledger/shared-goal-authority-state-provider-v0/2026-09-28-retirement-cadence.zh-CN.md index c53557a252..e0462baffb 100644 --- a/docs/architecture/rfcs/ledger/shared-goal-authority-state-provider-v0/2026-09-28-retirement-cadence.zh-CN.md +++ b/docs/architecture/rfcs/ledger/shared-goal-authority-state-provider-v0/2026-09-28-retirement-cadence.zh-CN.md @@ -354,8 +354,33 @@ Quota 观察复用既有 should-run 摘要：捕获的单 Goal 行序列化由 1 另一项 148 秒隔离演练通过新进程为两种 provider 各追加 12 笔提交，跨越 checkpoint， 逐轮验证原回执重放、变更意图拒绝及 projection／hash 一致性。它证明这段有界存储 流程，**不代表** Host 执行、活跃 Goal 采用或 D2 的十天 soak 已完成。本次不改变活跃 -authority、发布默认值或旧 writer 删除决定。B 仍缺持续负载／平台／容量证据；C 仍需 -consumer／新建入口及受支持升级验收。 +authority、发布默认值或旧 writer 删除决定。C 仍需 consumer／新建入口及受支持升级验收。 + +原封不动的 `matched-64k` capacity runner 在 `5f51559dc` 源码上完成独立的 +10k／100k 历史：macOS arm64、Node 24.21.0／SQLite 3.53.4、WAL/FULL， +两轴使用相同且非空的 1,341 文件公共扫描根。正式 ledger 为 +**13 passed／1 failed／11 missing**，进程退出 1，`full_d2_qualified=false`。 +部分 p95 测量如下： + +| 测量 | 10k | 100k | 冻结预算／结果 | +| --- | ---: | ---: | --- | +| Warm head | 2.360 ms | 4.860 ms | 增长 2.059 倍 > 2 倍：**失败**；绝对值 < 50 ms 通过 | +| Commit | 4.654 ms | 7.022 ms | 增长 1.509 倍、绝对值 < 100 ms 通过 | +| Receipt | 5.043 ms | 6.952 ms | 增长 1.379 倍、绝对值 < 50 ms 通过 | +| 新进程 CLI status | 1,133.177 ms | 1,143.820 ms | < 2,000 ms 通过 | +| 新进程 CLI mutation | 671.668 ms | 865.628 ms | 增量 193.960 ms < 200 ms 通过 | + +真实 CLI 操作、head／receipt 校验和临时存储清理均完成，没有正确性断言失败。 +这两条存储轴不证明完整八 agent／四 writer 负载、稳定 RSS、大历史恢复、consumer +lag、升级／回退、十日 soak、受支持 runtime 矩阵、promotion、1 MiB payload、 +300k 余量或 60 秒 burst；没有清空 OS 文件缓存。 +[D2 issue #4224](https://github.com/loopx-project/loopx/issues/4224) 已记录 soak 启动， +须先与既有负责人核实完成报告及其对当前候选的适用性，再决定是否授权替代运行。 + +下一项有界工作先区分连续性查询、proof／codec、连接与 Host 成本，再选择修复； +根因尚未证明。保留原失败和预算，不能删历史 hole／摘要／receipt 校验、缩短负载 +或反复重跑到绿；head 绝对值通过不能抵消增长预算失败。B 保持 open，这项单源码 +结果不证明发行回归，也不改变 provider 默认值或当前 File 对照要求。 ### 合同健康检查的权威归属