Repository navigation
task: B01 — 旧 run/evaluate 整链迁入最终归属 #414
Description
已确认实施方案：用户于 2026-11-12 11:35–11:36（北京时间）确认综合推荐，并要求快速、干净切换。
- 目的：首次创建
benchmark/，把旧run链按职责拆完：规则进 B，执行事实留 H，组合进 Apps legacy。并删除该链上的 H→model 以及除进程域之外的 H→infra 边。 - 迁入或删除：
harness/evaluation.py的evaluate/_summary进 B；runner不再内嵌判分与写报告，改由 Apps legacy 组合；scenarios/external的执行留 H、metric 校验进 B；reporting.write_report进 Apps；collect_samples进 B 采样（D008 未列此函数，本 PR 须先补 owner 再搬）。留在 H 的写入改用 H 自有原子 JSON（与infra.artifacts字节等价）。model/protocols/metrics可先由 B 导入，物理迁移留给后续 validators 切片。旧 harness 路径不留会形成 H→B/Apps 的壳；根evaluation.py/execution.py/reporting.py可暂留精确转发，到 R01 再删。 - 前置：H02、Publish standardized Apps run output #397；task: Apps 应用层详细设计与迁移计划（CLI / Python / Plugin） #371 接受
apps/legacy/路径。 - 验收：删除 evaluation/external 的 model 边，以及 runner/external/media 对 artifacts/targets 的边；旧
run退出码 0/1/2、incomplete 优先、同一夹具下evaluation.json字节不变；absent: benchmark/消失；overall 仍为 false。 - 明确不做：skill-eval 族、B 的
prepare/assess生产实现、model/validators迁移、根转发退役、CLI 迁移。
合入后 overall isolation_complete 必须仍为 false。
关联架构总题 #387、Apps 设计 #371；#389 是当前清单与 observability 来源，#398 为已合入 H01，#397 保持纯 Apps 标准输出且只读不扩范围。
Activity
amy-why-3459 commentedon Nov 12, 2026
谢谢，B01 的拆分思路清楚。我对照了当前 main（
e2f4cd1e）上的相关文件，补充几点：harness/runner.py现有的边（L11-16）：infra.artifacts.atomic_write_json、infra.targets.repository_identity、.evaluation.evaluate、.reporting.write_report、.scenarios.external。正文覆盖了 evaluate、reporting 和 artifacts。repository_identity归谁（H 自备，还是由 Apps legacy 组合时传入）建议在 PR 里写明，因为验收要求"runner 对 artifacts/targets 的边"都消失。harness/evaluation.py还导入了..claim.Verdict（L5）和..runspec.RunSpec（L7）。evaluate/_summary搬进 B 后，B→claim/runspec这两条边是否合规，请对照 rfc: K/H/B 与 Apps 公开接口详细设计（最小交接与模块边界） #387 的目标图确认，必要时一并登记。evaluation.json字节不变"和"退出码 0/1/2、incomplete 优先"，建议各配一个迁移前后对比的夹具测试，至少覆盖 pass、fail 和 incomplete 三种情况。status:proposed，Publish standardized Apps run output #397 仍是 Draft，main 上还没有apps/。正文已把"task: Apps 应用层详细设计与迁移计划（CLI / Python / Plugin） #371 接受apps/legacy/路径"列为前置，这一点很好，请在开工前确认。Merge branch 'main' into impl/b01-run-evaluate
test: pin sampling._sha256 to infra.artifacts.sha256
docs: record RSI-master D008 B01 ownership decisions
docs: cite delegated D008 decisions and clear review nits
benchmark: present the B01 tree as a single commit on current main
Dong1017 commentedon Nov 15, 2026
@amy-why-3459 @zengchuang-hw — under the delegated architecture-coordination role, I adopt the joint five-entry D008 record for these two governing issues:
B01 delivers its implementation plus this shared decision metadata; OBS retains its implementation. This records architecture adoption, not independent implementation acceptance. Both issues retain their evidence requirements;
isolation_completeremains false.benchmark: present the B01 tree as a single commit on current main (#422
Dong1017 commentedon Nov 15, 2026
Closing the bounded B01 migration slice delivered by PR #422, merged as
9ede1835.The legacy run/evaluate chain is assigned to Benchmark, Harness facts and Apps composition, with compatibility retained. Evidence includes hosted checks and the PR's exact-head Linux regression. The shared D008 record was adopted on #414 and #413.
A-entry, R01 and #410's formal main-only replay remain separate. Overall
isolation_completeremains false.