stratagate-dsh 0.3.26 → 0.3.27
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/CHANGELOG.md +679 -735
- package/LICENSE +21 -21
- package/dist/client.js +5 -6
- package/dist/index.js.map +1 -1
- package/dist/plugin.js +623 -164
- package/dist/plugin.js.map +1 -1
- package/docs/ARCHITECTURE.md +4 -0
- package/docs/DSH.md +7 -3
- package/docs/DSH.zh-CN.md +7 -3
- package/docs/benchmarks/issue-128-windows-review.json +104 -104
- package/docs/benchmarks/issue-128-windows.json +103 -103
- package/docs/feedback.md +50 -50
- package/docs/issue-128-lease-cache-review-followup.md +55 -55
- package/docs/issue-128-review-followup.md +58 -58
- package/docs/knowledge-graph.md +38 -0
- package/docs/npm-release.md +52 -52
- package/package.json +13 -13
- package/screenshots.json +8 -8
- package/docs/validation/frontend-feedback-20261021/background-refresh-failed.png +0 -0
- package/docs/validation/frontend-feedback-20261021/browser-results.json +0 -334
- package/docs/validation/frontend-feedback-20261021/first-read-graph.png +0 -0
- package/docs/validation/frontend-feedback-20261021/first-read-profile.png +0 -0
- package/docs/validation/frontend-feedback-20261021/first-read-short.png +0 -0
- package/docs/validation/frontend-feedback-20261021/profile-conflict.png +0 -0
- package/docs/validation/frontend-feedback-20261021/profile-save-regression.json +0 -466
- package/docs/validation/frontend-feedback-20261021/report.json +0 -99
- package/docs/validation/frontend-feedback-20261021/settings-saved-read-failed.png +0 -0
- package/docs/validation/frontend-feedback-20261021/settings-unknown.png +0 -0
- package/docs/validation/frontend-feedback-20261021/unit-results.json +0 -1
- package/docs/validation/issue119.md +0 -41
- package/docs/validation/workspace-cleanup-20261020.md +0 -36
package/docs/ARCHITECTURE.md
CHANGED
|
@@ -4,6 +4,10 @@ StrataGate separates source preservation, derived memory, retrieval control, and
|
|
|
4
4
|
|
|
5
5
|
## System boundaries
|
|
6
6
|
|
|
7
|
+
Knowledge Graph processing is optional and defaults off for new and upgraded databases. DSH persists the installation policy in its existing settings table; Core persists namespace defaults using additive columns without deleting graph data. A durable epoch fences claims, provider retries, lease renewals and late commits. Event ingestion can still commit when Graph is disabled during extraction, but cannot enqueue new Graph work. Ordinary retrieval, topics, Profile and Dreaming remain independent.
|
|
8
|
+
|
|
9
|
+
Historical Graph processing requires a preview and explicit confirmation of a finite Event ID list. Existing bounded batches, leases and retry budgets remain in effect; pause/resume does not reset spent attempts. Read-only admin browsing does not migrate the database. Older binaries can open the database but do not honor this switch, so downgrades may resume Graph work. See [Graph policy details](knowledge-graph.md).
|
|
10
|
+
|
|
7
11
|
```mermaid
|
|
8
12
|
flowchart TB
|
|
9
13
|
subgraph Source["Source layer"]
|
package/docs/DSH.md
CHANGED
|
@@ -18,6 +18,10 @@ The plugin adapts DSH session events to the existing StrataGate memory engine; i
|
|
|
18
18
|
|
|
19
19
|
## How it is designed
|
|
20
20
|
|
|
21
|
+
### Optional Knowledge Graph, off by default
|
|
22
|
+
|
|
23
|
+
New installations hide the Graph page until enabled in **More → Advanced settings → Memory configuration**. Existing installations upgrade with Graph processing off, retained data and a disabled Graph view; a one-time settings notice explains the additional model and token cost. Enabling or resuming previews pending work and requires confirmation for a finite historical list, processed in batches with pause/resume and existing retry limits. Events, directory, ordinary retrieval, Profile and Dreaming continue independently. Older versions do not recognize this switch and may restart Graph processing after a downgrade. See [policy and compatibility details](knowledge-graph.md).
|
|
24
|
+
|
|
21
25
|
### Directory Dreaming (development)
|
|
22
26
|
|
|
23
27
|
Click **🌙 Dreaming** in the long-term memory directory to let a dedicated Agent work through existing chapters, sections, Event membership, overviews, and source-backed experience. The panel shows an animated moon, completed-batch progress, pause, resume, and terminate controls. Closing settings leaves the task running. After a host restart, resume its durable checkpoint from the directory; committed edits are retained.
|
|
@@ -229,7 +233,7 @@ Open DSH Settings and select **StrataGate-AgentMemory**. The page provides:
|
|
|
229
233
|
- manual Block expansion and a two-step external-memory import flow;
|
|
230
234
|
- a Usage Audit chain from a recorded answer turn, through the Evidence Gate verdict and selected memories, back to source messages.
|
|
231
235
|
|
|
232
|
-
Events, graph facts, and source messages cannot be edited, deleted, or approved in the UI.
|
|
236
|
+
Events, graph facts, and source messages cannot be edited, deleted, or approved in the UI. Advanced Settings separates StrataGate installation-level controls from host configuration and current-workspace diagnostics. The installation-level controls change completed turns per Block, the global Block decay coefficient λ, and the agent-memory retrieval weight. They apply to every workspace in the current StrataGate installation, become defaults for future workspaces, and are persisted in the current memory database; they do not sync automatically across devices or installations. When the Block size changes, the UI explains its relationship to λ and suggests a value that preserves decay per conversation turn; the user decides whether to adopt it. Existing sealed Blocks are never repartitioned. The structured reasoning effort is a host setting in the current DSH profile and is saved by the host for subsequent memory-model calls. Schema, workspace identity, turn counts, and diagnostics describe the selected workspace.
|
|
233
237
|
|
|
234
238
|
The primary Persistent Profile tab shows eleven fixed fields in compact groups, one field at a time. An existing `homeCity` value adds a separate editable legacy row. It refreshes about every five seconds while visible, preserves an open draft, and prevents a stale same-field save from overwriting a concurrent change. The same installation-wide Profile is used by `memory_profile_update`, across sessions and namespaces. The preferred answer language applies only to the final answer; the independent visible reasoning language applies only to reasoning/thinking text exposed by the host UI when supported, never to hidden internal reasoning. An empty visible reasoning language adds no extra requirement. Each saved field takes effect on the next model call; empty fields are omitted. The stored Profile has twelve fields with Unicode code-point limits of 100/100/100/100/200/100/100/1000/1000/1500/1000/1200 characters, with a 6000-character total budget and a 4800-character maintenance trigger. The SQLite database records each Settings, tool, and maintenance change with its previous value and source. Maintenance runs when 24 hours have passed since the first nonempty write or the last successful run, or when a capacity threshold is reached; it sees only the current Profile and may compress wording without adding facts or semantically rewriting the seven short fields. Event extraction and Graph projection never write Profile fields.
|
|
235
239
|
|
|
@@ -266,7 +270,7 @@ Topic Projection prioritizes complete chapter and section assignments for every
|
|
|
266
270
|
|
|
267
271
|
Only explicit `finish=max-tokens` splits a failed batch in half, with at most four levels and three attempts per singleton. Each historical child still consumes the shared Bootstrap allowance, without trimming Event evidence. Other failures keep bounded retry and backoff. After automatic attempts are exhausted, the directory retry action (`POST /api/stratagate/topics/retry`, with namespace, jobId and expectedRevision) or `POST /api/stratagate/jobs/retry?namespace=...&kind=topic-projection&jobId=...` queues a fresh job ID. Both paths enforce source versions, reject old results, and persist across restart. Recovery Version 2 reuses the existing persistent writer-initialization mechanism to queue one new cycle of at most three task attempts for current terminal Topic failures with unfinished work that match overview validation, legacy draft requirements, Provider failure or the original `invalid-output` signature. Unrelated worker, timeout and content validation failures retain their attempt budgets. This cycle consumes the existing historical allowance. Successful work, stale sources and retired history are excluded. The recovery marker and queued jobs commit together; restarts, refreshes and repeated initialization never reset attempts again. If the new cycle fails, partial progress, error details and manual retry remain available. Only fully validated current work reports `completed`; exhausted outstanding work reports `failed`.
|
|
268
272
|
|
|
269
|
-
`blockTurnSize` and `
|
|
273
|
+
`blockTurnSize`, `blockDecayLambda`, and `agentMemoryRetrievalWeight` are initial fallbacks. Once changed in **Advanced Settings**, installation-level values persisted in the current memory database take precedence for every workspace and future workspace in that installation. They do not sync automatically across devices or installations. λ defaults to `0.3`; smaller values forget more slowly and consume more tokens, and values above `0.4` are not recommended.
|
|
270
274
|
|
|
271
275
|
`project` derives a stable namespace from the normalized session working directory. `session` isolates every DSH session. `global` shares one namespace.
|
|
272
276
|
|
|
@@ -298,7 +302,7 @@ On DSH `0.2.0`, the host-owned `dsh-session-format`, `dsh-session-format-catalog
|
|
|
298
302
|
|
|
299
303
|
DSH core packages are host-provided optional peers. After upgrading a profile that once installed those peers locally, run `dsh plugin --profile <name> exec stratagate-dsh-repair` before starting it. The command moves only the known DSH runtime packages into `.stratagate-runtime-backups` inside that profile and writes a recovery receipt; it does not delete them or touch session data. At startup, a bootstrap resolver also forces StrataGate's own imports through the installation-owned module fallback, then verifies the resolved DSH core family and stops with an actionable error for an unknown host family. On supported DSH `0.1.5`, `0.1.6`, `0.1.7`, and `0.2.0` hosts, v0 sessions containing the retired `stratagate/memory-citations` event are copied to a validated v1 generation before the host continues its migration chain. The original v0 log is never modified; a `stratagate-legacy-citations-v1.json` receipt records source/target hashes and recovery instructions.
|
|
300
304
|
|
|
301
|
-
DSH 0.1.7 and 0.2.0 store plugin configuration in the profile and present live controls through config forms. StrataGate's chat settings use those forms on 0.1.7 and 0.2.0 and the older settings service on earlier supported hosts.
|
|
305
|
+
DSH 0.1.7 and 0.2.0 store plugin configuration in the current profile and present live controls through config forms. StrataGate's chat settings use those forms on 0.1.7 and 0.2.0 and the older settings service on earlier supported hosts. The structured reasoning effort is shared by workspaces in that profile; changing it takes effect on subsequent memory-model calls without a plugin restart.
|
|
302
306
|
|
|
303
307
|
DSH 0.1.6's official DeepSeek profile may enable Session Log request metadata by default, so the Host can include original Session Events in the `dsh_session_log` request field. This metadata is outside the model messages, system prompt, and tool schema and does not indicate that StrataGate compression was bypassed. StrataGate does not alter this DSH-wide setting; users who want it disabled should set `session-log-deepseek.enabled=false` through DSH configuration.
|
|
304
308
|
|
package/docs/DSH.zh-CN.md
CHANGED
|
@@ -18,6 +18,10 @@
|
|
|
18
18
|
|
|
19
19
|
## 它是怎么设计的
|
|
20
20
|
|
|
21
|
+
### 知识图谱可选,默认关闭
|
|
22
|
+
|
|
23
|
+
新用户默认隐藏图谱页;存量用户升级后关闭处理、保留全部数据和灰色入口,首次进入记忆设置显示一次说明。在 **更多 → 高级设置 → 记忆配置** 开启时先统计待处理数量,确认后按有限清单分批恢复,可暂停和继续。事件卡、目录、普通检索、Profile 和 Dreaming 独立工作。旧版本不认识此开关,降级运行可能重新启用图谱处理。详见 [开关、迁移和兼容边界](knowledge-graph.md)。
|
|
24
|
+
|
|
21
25
|
### 目录 Dreaming(开发版)
|
|
22
26
|
|
|
23
27
|
在长期记忆的目录页点击 **🌙 Dreaming**,由独立 Agent 分批整理已有章节、小节、Event 归属、总览与有来源的经验。运行时显示做梦动画和真实批次进度,可暂停、继续或终止。关闭设置页不取消任务;宿主重启后从目录页继续已保存的断点,已提交修改保留。
|
|
@@ -192,7 +196,7 @@ ID、`blockId`、角色、轮次和有界摘录。`narrative`、`quotes`、来
|
|
|
192
196
|
- 手动展开 Block,以及分两步导入其他 AI 的记忆;
|
|
193
197
|
- Usage Audit(使用审计)链路:从已记录的回答轮次出发,经由 Evidence Gate 的判断与选中的记忆,追溯到来源消息。
|
|
194
198
|
|
|
195
|
-
界面不允许直接编辑、删除或批准 Event、图谱事实和来源消息。用户可以手动展开 Block、导入其他 AI
|
|
199
|
+
界面不允许直接编辑、删除或批准 Event、图谱事实和来源消息。用户可以手动展开 Block、导入其他 AI 的记忆,以及在“高级设置”的“StrataGate 安装级设置”中修改每个 Block 包含的完整对话轮数、Block 衰减系数 λ 和主动记忆检索占比。这三项适用于当前 StrataGate 安装中的所有工作区,也作为新工作区的默认设置;保存到当前记忆数据库,不会自动跨设备或跨安装同步。修改轮数时,界面会解释两者关系并给出保持单位对话衰减速度的建议 λ,是否采用由用户决定。保存后设置立即应用到所有已有工作区,并在重启后保持;已封存 Block 不会重新切分。结构化推理档位属于当前 DSH profile 的宿主配置,由宿主立即保存并作用于后续记忆处理调用。Schema、工作区名称、内部空间 ID、处理轮次以及运行诊断随当前工作区选择器切换,仅描述当前工作区。
|
|
196
200
|
|
|
197
201
|
顶部“常驻画像”页面默认显示十一项字段;旧常驻城市有值时额外保留一项编辑入口。页面按身份与语言、交互偏好、常用地址、关于用户、其他分组,默认紧凑显示;编辑时只展开一个字段。“默认回答语言”仅用于最终回复,“思考过程语言”仅用于宿主界面向用户展示的 reasoning/thinking 文本(若支持),两项互不推导;后者为空时不额外要求思考语言,也不控制隐藏的内部推理。页面可见时约每 5 秒读取最新画像,隐藏或离开时停止,并保留正在编辑的草稿。同一字段被其他来源修改时会提示冲突,保存不会覆盖新值。Settings 与 `memory_profile_update` 修改同一份安装级画像,跨会话、跨命名空间生效;保存后下一次模型调用立即看到新值,空字段不注入。存储结构有十二项字段,按 Unicode 码点计数,上限依次为 100、100、100、100、200、100、100、1000、1000、1500、1000、1200 字符,总预算 6000 字符;总容量 4800 字符触发后台整理。SQLite 会记录 Settings、工具及后台整理的每次变更、旧值和来源。后台整理在首次写入非空画像或上次成功整理满 24 小时后,或达到容量阈值时执行;只读取当前画像,只压缩表达,不新增事实,也不语义改写七个短字段。Event 提取和 Graph 投影都不会写入画像。
|
|
198
202
|
|
|
@@ -229,7 +233,7 @@ Topic Projection 优先将全部本批 Event 完整分配到章节和小节。
|
|
|
229
233
|
|
|
230
234
|
只有明确的 `finish=max-tokens` 才会二分失败批次,最多四层,单条批次最多尝试三次;每个历史子批次仍受共用 Bootstrap 额度约束,不裁剪 Event 证据。其他错误维持原有有限重试与退避。自动尝试耗尽后,可通过目录“重新整理这 N 条”(`POST /api/stratagate/topics/retry`,携带 namespace、jobId 和 expectedRevision),或 `POST /api/stratagate/jobs/retry?namespace=...&kind=topic-projection&jobId=...` 重新排队。两条路径均开启新 jobId 周期,检查来源版本并拒绝旧结果,状态在重启后保留。Recovery Version 2 复用现有持久化 writer 初始化机制,仅为命中概要校验、旧草稿要求、Provider 失败或原始 `invalid-output` 签名,且当前有效、仍有未完成工作的 Topic 终态失败任务自动开启一次新的最多三次任务尝试,仍消耗已有历史额度。无关 worker、超时及内容校验失败保留原尝试额度。已成功、来源过期及失效的历史记录不重跑;恢复标记与新任务一起提交,重启、刷新和重复初始化不再次清零。本轮再次失败后保留部分完成进度、错误详情和人工重试入口。只有当前应完成工作全部通过校验才返回 `completed`,仅剩终态失败时返回 `failed`。
|
|
231
235
|
|
|
232
|
-
配置文件中的 `blockTurnSize` 和 `
|
|
236
|
+
配置文件中的 `blockTurnSize`、`blockDecayLambda` 和 `agentMemoryRetrievalWeight` 是初始后备值;一旦在“高级设置”中修改,当前记忆数据库中的安装级值优先生效,并供所有工作区及新工作区使用。它们不会自动跨设备或跨安装同步。λ 默认值为 `0.3`;数字越小,记忆遗忘越慢、消耗 token 越多,不建议大于 `0.4`。
|
|
233
237
|
|
|
234
238
|
`project` 会根据规范化后的会话工作目录生成稳定的命名空间;`session` 会隔离每个 DSH 会话;`global` 则让所有会话共享同一个命名空间。
|
|
235
239
|
|
|
@@ -259,7 +263,7 @@ Topic Projection 优先将全部本批 Event 完整分配到章节和小节。
|
|
|
259
263
|
|
|
260
264
|
DSH 核心包现在都是由宿主提供的可选 peer。若某个 profile 曾在本地安装这些 peer,升级后、启动前运行 `dsh plugin --profile <名称> exec stratagate-dsh-repair`。该命令只会把已知 DSH 运行时包移动到 profile 内的 `.stratagate-runtime-backups` 并写入恢复收据,不会删除包或触碰会话数据。启动时 bootstrap 还会把 StrataGate 自身的 DSH 导入定向到宿主维护的共享模块回退目录,再检查实际解析出的 DSH 核心族;未知宿主会立即给出明确错误并停止,而不是让正文区域空白。对于受支持的 DSH `0.1.5`、`0.1.6`、`0.1.7` 和 `0.2.0`,含已废弃 `stratagate/memory-citations` 事件的 v0 会话会先复制成经过校验的 v1 代际,再由宿主继续其迁移链;原始 v0 日志不会被修改,并写入 `stratagate-legacy-citations-v1.json` 记录源/目标哈希和恢复说明。
|
|
261
265
|
|
|
262
|
-
DSH 0.1.7 和 0.2.0
|
|
266
|
+
DSH 0.1.7 和 0.2.0 将插件配置保存在当前 profile 中,并通过配置表单提供实时控制。StrataGate 聊天界面中的设置在 0.1.7 和 0.2.0 使用配置表单,在之前受支持的宿主上继续使用旧设置服务。结构化推理档位是当前 profile 的宿主配置,由该 profile 下的工作区共享;调整后,后续记忆模型调用会直接使用新值,无需重启插件。
|
|
263
267
|
|
|
264
268
|
DSH 0.1.6 的官方 DeepSeek profile 默认可能启用 Session Log 请求元数据,因此宿主可能把原始 Session Event 放入 `dsh_session_log` 请求字段。该元数据不属于模型 messages、system prompt 或 tool schema,不能据此判断 StrataGate 压缩失效。StrataGate 不会修改这一 DSH 全局设置;需要关闭的用户应通过 DSH 配置设置 `session-log-deepseek.enabled=false`。
|
|
265
269
|
|
|
@@ -1,104 +1,104 @@
|
|
|
1
|
-
{
|
|
2
|
-
"baselineSha": "e954eaf2f2cbbef4bb6226a2ebb35191424868c2",
|
|
3
|
-
"sourceSha": "415096f19d09d9bcc29c44634c7ae78fd6c2f52d",
|
|
4
|
-
"capturedAt": "2026-10-10T11:22:46.467Z",
|
|
5
|
-
"scope": "Synthetic DSH runtime on Windows; not the user database or full DSH UI process.",
|
|
6
|
-
"scenarios": [
|
|
7
|
-
{
|
|
8
|
-
"name": "small",
|
|
9
|
-
"namespaces": 100,
|
|
10
|
-
"rawUserCharsPerNamespace": 8192,
|
|
11
|
-
"databaseBytes": 8941568,
|
|
12
|
-
"rounds": 3,
|
|
13
|
-
"intervalMs": 3000,
|
|
14
|
-
"measurements": {
|
|
15
|
-
"before": {
|
|
16
|
-
"pollsMs": [
|
|
17
|
-
2035.9983000000002,
|
|
18
|
-
2204.014,
|
|
19
|
-
2077.0357000000004
|
|
20
|
-
],
|
|
21
|
-
"meanPollMs": 2105.682666666667,
|
|
22
|
-
"fullMemoryLoads": 300,
|
|
23
|
-
"cpuMs": 3015,
|
|
24
|
-
"wallMs": 15335.1882,
|
|
25
|
-
"processCpuPercent": 19.660665136147465,
|
|
26
|
-
"rssBefore": 56864768,
|
|
27
|
-
"rssAfter": 73207808,
|
|
28
|
-
"peakRss": 74285056,
|
|
29
|
-
"node": "v24.19.0",
|
|
30
|
-
"platform": "win32",
|
|
31
|
-
"arch": "x64",
|
|
32
|
-
"cpuMeasurementNote": "A zero CPU delta means below Windows counter resolution in this sample, not zero actual CPU work."
|
|
33
|
-
},
|
|
34
|
-
"after": {
|
|
35
|
-
"pollsMs": [
|
|
36
|
-
32.8809,
|
|
37
|
-
5.330300000000079,
|
|
38
|
-
2.298700000000281
|
|
39
|
-
],
|
|
40
|
-
"meanPollMs": 13.503300000000118,
|
|
41
|
-
"fullMemoryLoads": 0,
|
|
42
|
-
"cpuMs": 31,
|
|
43
|
-
"wallMs": 9071.7334,
|
|
44
|
-
"processCpuPercent": 0.3417208005693818,
|
|
45
|
-
"rssBefore": 56606720,
|
|
46
|
-
"rssAfter": 54337536,
|
|
47
|
-
"peakRss": 57950208,
|
|
48
|
-
"node": "v24.19.0",
|
|
49
|
-
"platform": "win32",
|
|
50
|
-
"arch": "x64",
|
|
51
|
-
"cpuMeasurementNote": "A zero CPU delta means below Windows counter resolution in this sample, not zero actual CPU work."
|
|
52
|
-
}
|
|
53
|
-
}
|
|
54
|
-
},
|
|
55
|
-
{
|
|
56
|
-
"name": "large",
|
|
57
|
-
"namespaces": 100,
|
|
58
|
-
"rawUserCharsPerNamespace": 1572864,
|
|
59
|
-
"databaseBytes": 1519824896,
|
|
60
|
-
"rounds": 3,
|
|
61
|
-
"intervalMs": 3000,
|
|
62
|
-
"measurements": {
|
|
63
|
-
"before": {
|
|
64
|
-
"pollsMs": [
|
|
65
|
-
12951.9511,
|
|
66
|
-
8731.1299,
|
|
67
|
-
9476.834900000005
|
|
68
|
-
],
|
|
69
|
-
"meanPollMs": 10386.638633333334,
|
|
70
|
-
"fullMemoryLoads": 300,
|
|
71
|
-
"cpuMs": 24844,
|
|
72
|
-
"wallMs": 40200.4599,
|
|
73
|
-
"processCpuPercent": 61.80028801113292,
|
|
74
|
-
"rssBefore": 56283136,
|
|
75
|
-
"rssAfter": 157401088,
|
|
76
|
-
"peakRss": 332111872,
|
|
77
|
-
"node": "v24.19.0",
|
|
78
|
-
"platform": "win32",
|
|
79
|
-
"arch": "x64",
|
|
80
|
-
"cpuMeasurementNote": "A zero CPU delta means below Windows counter resolution in this sample, not zero actual CPU work."
|
|
81
|
-
},
|
|
82
|
-
"after": {
|
|
83
|
-
"pollsMs": [
|
|
84
|
-
29.53,
|
|
85
|
-
3.372899999999845,
|
|
86
|
-
2.309900000000198
|
|
87
|
-
],
|
|
88
|
-
"meanPollMs": 11.737600000000015,
|
|
89
|
-
"fullMemoryLoads": 0,
|
|
90
|
-
"cpuMs": 109,
|
|
91
|
-
"wallMs": 9061.4375,
|
|
92
|
-
"processCpuPercent": 1.2028996503038287,
|
|
93
|
-
"rssBefore": 52092928,
|
|
94
|
-
"rssAfter": 51056640,
|
|
95
|
-
"peakRss": 54579200,
|
|
96
|
-
"node": "v24.19.0",
|
|
97
|
-
"platform": "win32",
|
|
98
|
-
"arch": "x64",
|
|
99
|
-
"cpuMeasurementNote": "A zero CPU delta means below Windows counter resolution in this sample, not zero actual CPU work."
|
|
100
|
-
}
|
|
101
|
-
}
|
|
102
|
-
}
|
|
103
|
-
]
|
|
104
|
-
}
|
|
1
|
+
{
|
|
2
|
+
"baselineSha": "e954eaf2f2cbbef4bb6226a2ebb35191424868c2",
|
|
3
|
+
"sourceSha": "415096f19d09d9bcc29c44634c7ae78fd6c2f52d",
|
|
4
|
+
"capturedAt": "2026-10-10T11:22:46.467Z",
|
|
5
|
+
"scope": "Synthetic DSH runtime on Windows; not the user database or full DSH UI process.",
|
|
6
|
+
"scenarios": [
|
|
7
|
+
{
|
|
8
|
+
"name": "small",
|
|
9
|
+
"namespaces": 100,
|
|
10
|
+
"rawUserCharsPerNamespace": 8192,
|
|
11
|
+
"databaseBytes": 8941568,
|
|
12
|
+
"rounds": 3,
|
|
13
|
+
"intervalMs": 3000,
|
|
14
|
+
"measurements": {
|
|
15
|
+
"before": {
|
|
16
|
+
"pollsMs": [
|
|
17
|
+
2035.9983000000002,
|
|
18
|
+
2204.014,
|
|
19
|
+
2077.0357000000004
|
|
20
|
+
],
|
|
21
|
+
"meanPollMs": 2105.682666666667,
|
|
22
|
+
"fullMemoryLoads": 300,
|
|
23
|
+
"cpuMs": 3015,
|
|
24
|
+
"wallMs": 15335.1882,
|
|
25
|
+
"processCpuPercent": 19.660665136147465,
|
|
26
|
+
"rssBefore": 56864768,
|
|
27
|
+
"rssAfter": 73207808,
|
|
28
|
+
"peakRss": 74285056,
|
|
29
|
+
"node": "v24.19.0",
|
|
30
|
+
"platform": "win32",
|
|
31
|
+
"arch": "x64",
|
|
32
|
+
"cpuMeasurementNote": "A zero CPU delta means below Windows counter resolution in this sample, not zero actual CPU work."
|
|
33
|
+
},
|
|
34
|
+
"after": {
|
|
35
|
+
"pollsMs": [
|
|
36
|
+
32.8809,
|
|
37
|
+
5.330300000000079,
|
|
38
|
+
2.298700000000281
|
|
39
|
+
],
|
|
40
|
+
"meanPollMs": 13.503300000000118,
|
|
41
|
+
"fullMemoryLoads": 0,
|
|
42
|
+
"cpuMs": 31,
|
|
43
|
+
"wallMs": 9071.7334,
|
|
44
|
+
"processCpuPercent": 0.3417208005693818,
|
|
45
|
+
"rssBefore": 56606720,
|
|
46
|
+
"rssAfter": 54337536,
|
|
47
|
+
"peakRss": 57950208,
|
|
48
|
+
"node": "v24.19.0",
|
|
49
|
+
"platform": "win32",
|
|
50
|
+
"arch": "x64",
|
|
51
|
+
"cpuMeasurementNote": "A zero CPU delta means below Windows counter resolution in this sample, not zero actual CPU work."
|
|
52
|
+
}
|
|
53
|
+
}
|
|
54
|
+
},
|
|
55
|
+
{
|
|
56
|
+
"name": "large",
|
|
57
|
+
"namespaces": 100,
|
|
58
|
+
"rawUserCharsPerNamespace": 1572864,
|
|
59
|
+
"databaseBytes": 1519824896,
|
|
60
|
+
"rounds": 3,
|
|
61
|
+
"intervalMs": 3000,
|
|
62
|
+
"measurements": {
|
|
63
|
+
"before": {
|
|
64
|
+
"pollsMs": [
|
|
65
|
+
12951.9511,
|
|
66
|
+
8731.1299,
|
|
67
|
+
9476.834900000005
|
|
68
|
+
],
|
|
69
|
+
"meanPollMs": 10386.638633333334,
|
|
70
|
+
"fullMemoryLoads": 300,
|
|
71
|
+
"cpuMs": 24844,
|
|
72
|
+
"wallMs": 40200.4599,
|
|
73
|
+
"processCpuPercent": 61.80028801113292,
|
|
74
|
+
"rssBefore": 56283136,
|
|
75
|
+
"rssAfter": 157401088,
|
|
76
|
+
"peakRss": 332111872,
|
|
77
|
+
"node": "v24.19.0",
|
|
78
|
+
"platform": "win32",
|
|
79
|
+
"arch": "x64",
|
|
80
|
+
"cpuMeasurementNote": "A zero CPU delta means below Windows counter resolution in this sample, not zero actual CPU work."
|
|
81
|
+
},
|
|
82
|
+
"after": {
|
|
83
|
+
"pollsMs": [
|
|
84
|
+
29.53,
|
|
85
|
+
3.372899999999845,
|
|
86
|
+
2.309900000000198
|
|
87
|
+
],
|
|
88
|
+
"meanPollMs": 11.737600000000015,
|
|
89
|
+
"fullMemoryLoads": 0,
|
|
90
|
+
"cpuMs": 109,
|
|
91
|
+
"wallMs": 9061.4375,
|
|
92
|
+
"processCpuPercent": 1.2028996503038287,
|
|
93
|
+
"rssBefore": 52092928,
|
|
94
|
+
"rssAfter": 51056640,
|
|
95
|
+
"peakRss": 54579200,
|
|
96
|
+
"node": "v24.19.0",
|
|
97
|
+
"platform": "win32",
|
|
98
|
+
"arch": "x64",
|
|
99
|
+
"cpuMeasurementNote": "A zero CPU delta means below Windows counter resolution in this sample, not zero actual CPU work."
|
|
100
|
+
}
|
|
101
|
+
}
|
|
102
|
+
}
|
|
103
|
+
]
|
|
104
|
+
}
|
|
@@ -1,103 +1,103 @@
|
|
|
1
|
-
{
|
|
2
|
-
"baselineSha": "e954eaf2f2cbbef4bb6226a2ebb35191424868c2",
|
|
3
|
-
"capturedAt": "2026-10-09T14:26:54.222Z",
|
|
4
|
-
"scope": "Synthetic DSH runtime on Windows; not the user database or full DSH UI process.",
|
|
5
|
-
"scenarios": [
|
|
6
|
-
{
|
|
7
|
-
"name": "small",
|
|
8
|
-
"namespaces": 100,
|
|
9
|
-
"rawUserCharsPerNamespace": 8192,
|
|
10
|
-
"databaseBytes": 8949760,
|
|
11
|
-
"rounds": 3,
|
|
12
|
-
"intervalMs": 3000,
|
|
13
|
-
"measurements": {
|
|
14
|
-
"before": {
|
|
15
|
-
"pollsMs": [
|
|
16
|
-
1559.2007,
|
|
17
|
-
2646.6144999999997,
|
|
18
|
-
2287.6314999999995
|
|
19
|
-
],
|
|
20
|
-
"meanPollMs": 2164.482233333333,
|
|
21
|
-
"fullMemoryLoads": 300,
|
|
22
|
-
"cpuMs": 3547,
|
|
23
|
-
"wallMs": 15505.109400000001,
|
|
24
|
-
"processCpuPercent": 22.876330043824133,
|
|
25
|
-
"rssBefore": 56803328,
|
|
26
|
-
"rssAfter": 73904128,
|
|
27
|
-
"peakRss": 74313728,
|
|
28
|
-
"node": "v24.19.0",
|
|
29
|
-
"platform": "win32",
|
|
30
|
-
"arch": "x64",
|
|
31
|
-
"cpuMeasurementNote": "A zero CPU delta means below Windows counter resolution in this sample, not zero actual CPU work."
|
|
32
|
-
},
|
|
33
|
-
"after": {
|
|
34
|
-
"pollsMs": [
|
|
35
|
-
17.57029999999999,
|
|
36
|
-
3.0478000000002794,
|
|
37
|
-
4.158300000000054
|
|
38
|
-
],
|
|
39
|
-
"meanPollMs": 8.258800000000107,
|
|
40
|
-
"fullMemoryLoads": 0,
|
|
41
|
-
"cpuMs": 16,
|
|
42
|
-
"wallMs": 9039.4974,
|
|
43
|
-
"processCpuPercent": 0.17700099122767599,
|
|
44
|
-
"rssBefore": 56590336,
|
|
45
|
-
"rssAfter": 54194176,
|
|
46
|
-
"peakRss": 57184256,
|
|
47
|
-
"node": "v24.19.0",
|
|
48
|
-
"platform": "win32",
|
|
49
|
-
"arch": "x64",
|
|
50
|
-
"cpuMeasurementNote": "A zero CPU delta means below Windows counter resolution in this sample, not zero actual CPU work."
|
|
51
|
-
}
|
|
52
|
-
}
|
|
53
|
-
},
|
|
54
|
-
{
|
|
55
|
-
"name": "large",
|
|
56
|
-
"namespaces": 100,
|
|
57
|
-
"rawUserCharsPerNamespace": 1572864,
|
|
58
|
-
"databaseBytes": 1519833088,
|
|
59
|
-
"rounds": 3,
|
|
60
|
-
"intervalMs": 3000,
|
|
61
|
-
"measurements": {
|
|
62
|
-
"before": {
|
|
63
|
-
"pollsMs": [
|
|
64
|
-
13949.606099999999,
|
|
65
|
-
12921.8561,
|
|
66
|
-
12388.245900000002
|
|
67
|
-
],
|
|
68
|
-
"meanPollMs": 13086.569366666668,
|
|
69
|
-
"fullMemoryLoads": 300,
|
|
70
|
-
"cpuMs": 32203,
|
|
71
|
-
"wallMs": 48289.1154,
|
|
72
|
-
"processCpuPercent": 66.687906235698,
|
|
73
|
-
"rssBefore": 56573952,
|
|
74
|
-
"rssAfter": 180330496,
|
|
75
|
-
"peakRss": 284123136,
|
|
76
|
-
"node": "v24.19.0",
|
|
77
|
-
"platform": "win32",
|
|
78
|
-
"arch": "x64",
|
|
79
|
-
"cpuMeasurementNote": "A zero CPU delta means below Windows counter resolution in this sample, not zero actual CPU work."
|
|
80
|
-
},
|
|
81
|
-
"after": {
|
|
82
|
-
"pollsMs": [
|
|
83
|
-
40.84779999999998,
|
|
84
|
-
3.018500000000131,
|
|
85
|
-
3.4000999999998385
|
|
86
|
-
],
|
|
87
|
-
"meanPollMs": 15.75546666666665,
|
|
88
|
-
"fullMemoryLoads": 0,
|
|
89
|
-
"cpuMs": 63,
|
|
90
|
-
"wallMs": 9055.1278,
|
|
91
|
-
"processCpuPercent": 0.6957383859342107,
|
|
92
|
-
"rssBefore": 56807424,
|
|
93
|
-
"rssAfter": 54194176,
|
|
94
|
-
"peakRss": 57184256,
|
|
95
|
-
"node": "v24.19.0",
|
|
96
|
-
"platform": "win32",
|
|
97
|
-
"arch": "x64",
|
|
98
|
-
"cpuMeasurementNote": "A zero CPU delta means below Windows counter resolution in this sample, not zero actual CPU work."
|
|
99
|
-
}
|
|
100
|
-
}
|
|
101
|
-
}
|
|
102
|
-
]
|
|
103
|
-
}
|
|
1
|
+
{
|
|
2
|
+
"baselineSha": "e954eaf2f2cbbef4bb6226a2ebb35191424868c2",
|
|
3
|
+
"capturedAt": "2026-10-09T14:26:54.222Z",
|
|
4
|
+
"scope": "Synthetic DSH runtime on Windows; not the user database or full DSH UI process.",
|
|
5
|
+
"scenarios": [
|
|
6
|
+
{
|
|
7
|
+
"name": "small",
|
|
8
|
+
"namespaces": 100,
|
|
9
|
+
"rawUserCharsPerNamespace": 8192,
|
|
10
|
+
"databaseBytes": 8949760,
|
|
11
|
+
"rounds": 3,
|
|
12
|
+
"intervalMs": 3000,
|
|
13
|
+
"measurements": {
|
|
14
|
+
"before": {
|
|
15
|
+
"pollsMs": [
|
|
16
|
+
1559.2007,
|
|
17
|
+
2646.6144999999997,
|
|
18
|
+
2287.6314999999995
|
|
19
|
+
],
|
|
20
|
+
"meanPollMs": 2164.482233333333,
|
|
21
|
+
"fullMemoryLoads": 300,
|
|
22
|
+
"cpuMs": 3547,
|
|
23
|
+
"wallMs": 15505.109400000001,
|
|
24
|
+
"processCpuPercent": 22.876330043824133,
|
|
25
|
+
"rssBefore": 56803328,
|
|
26
|
+
"rssAfter": 73904128,
|
|
27
|
+
"peakRss": 74313728,
|
|
28
|
+
"node": "v24.19.0",
|
|
29
|
+
"platform": "win32",
|
|
30
|
+
"arch": "x64",
|
|
31
|
+
"cpuMeasurementNote": "A zero CPU delta means below Windows counter resolution in this sample, not zero actual CPU work."
|
|
32
|
+
},
|
|
33
|
+
"after": {
|
|
34
|
+
"pollsMs": [
|
|
35
|
+
17.57029999999999,
|
|
36
|
+
3.0478000000002794,
|
|
37
|
+
4.158300000000054
|
|
38
|
+
],
|
|
39
|
+
"meanPollMs": 8.258800000000107,
|
|
40
|
+
"fullMemoryLoads": 0,
|
|
41
|
+
"cpuMs": 16,
|
|
42
|
+
"wallMs": 9039.4974,
|
|
43
|
+
"processCpuPercent": 0.17700099122767599,
|
|
44
|
+
"rssBefore": 56590336,
|
|
45
|
+
"rssAfter": 54194176,
|
|
46
|
+
"peakRss": 57184256,
|
|
47
|
+
"node": "v24.19.0",
|
|
48
|
+
"platform": "win32",
|
|
49
|
+
"arch": "x64",
|
|
50
|
+
"cpuMeasurementNote": "A zero CPU delta means below Windows counter resolution in this sample, not zero actual CPU work."
|
|
51
|
+
}
|
|
52
|
+
}
|
|
53
|
+
},
|
|
54
|
+
{
|
|
55
|
+
"name": "large",
|
|
56
|
+
"namespaces": 100,
|
|
57
|
+
"rawUserCharsPerNamespace": 1572864,
|
|
58
|
+
"databaseBytes": 1519833088,
|
|
59
|
+
"rounds": 3,
|
|
60
|
+
"intervalMs": 3000,
|
|
61
|
+
"measurements": {
|
|
62
|
+
"before": {
|
|
63
|
+
"pollsMs": [
|
|
64
|
+
13949.606099999999,
|
|
65
|
+
12921.8561,
|
|
66
|
+
12388.245900000002
|
|
67
|
+
],
|
|
68
|
+
"meanPollMs": 13086.569366666668,
|
|
69
|
+
"fullMemoryLoads": 300,
|
|
70
|
+
"cpuMs": 32203,
|
|
71
|
+
"wallMs": 48289.1154,
|
|
72
|
+
"processCpuPercent": 66.687906235698,
|
|
73
|
+
"rssBefore": 56573952,
|
|
74
|
+
"rssAfter": 180330496,
|
|
75
|
+
"peakRss": 284123136,
|
|
76
|
+
"node": "v24.19.0",
|
|
77
|
+
"platform": "win32",
|
|
78
|
+
"arch": "x64",
|
|
79
|
+
"cpuMeasurementNote": "A zero CPU delta means below Windows counter resolution in this sample, not zero actual CPU work."
|
|
80
|
+
},
|
|
81
|
+
"after": {
|
|
82
|
+
"pollsMs": [
|
|
83
|
+
40.84779999999998,
|
|
84
|
+
3.018500000000131,
|
|
85
|
+
3.4000999999998385
|
|
86
|
+
],
|
|
87
|
+
"meanPollMs": 15.75546666666665,
|
|
88
|
+
"fullMemoryLoads": 0,
|
|
89
|
+
"cpuMs": 63,
|
|
90
|
+
"wallMs": 9055.1278,
|
|
91
|
+
"processCpuPercent": 0.6957383859342107,
|
|
92
|
+
"rssBefore": 56807424,
|
|
93
|
+
"rssAfter": 54194176,
|
|
94
|
+
"peakRss": 57184256,
|
|
95
|
+
"node": "v24.19.0",
|
|
96
|
+
"platform": "win32",
|
|
97
|
+
"arch": "x64",
|
|
98
|
+
"cpuMeasurementNote": "A zero CPU delta means below Windows counter resolution in this sample, not zero actual CPU work."
|
|
99
|
+
}
|
|
100
|
+
}
|
|
101
|
+
}
|
|
102
|
+
]
|
|
103
|
+
}
|