SQLite Web: gitea.db

id

Primary key. INTEGER NOT NULL

user_id

INTEGER

op_type

INTEGER

act_user_id

INTEGER

repo_id

INTEGER

comment_id

INTEGER

is_deleted

INTEGER NOT NULL (default 0

ref_name

TEXT

is_private

INTEGER NOT NULL (default 0

content

{"Commits":[{"Sha1":"05e3826e9e9e4dbbdd753252c2134b8494b08f2b","Message":"chore(reports): 工作台两份快照回绑 @ dfd4832\n\ndocs/runbook.md 在 workbench-ops-snapshot 的作用域里且是 error 级——改 §7 那段就会当场\n把 check:evidence 打红。经 reports:rebind 在 HEAD 的干净检出里重跑，两份同源快照一并带回。\n\nCo-Authored-By: Claude Opus 5 \u003cnoreply@anthropic.com\u003e\n","AuthorEmail":"hillao@juhailaoluodeMacBook-Pro.local","AuthorName":"juhailaoluo pro","CommitterEmail":"hillao@juhailaoluodeMacBook-Pro.local","CommitterName":"juhailaoluo pro","Timestamp":"2026-09-18T07:11:36-07:00"},{"Sha1":"dfd483287cee48db9791915cf33e037c4f758f6f","Message":"chore(reports): :68 修复波及的四份验收重跑回绑 @ 08c3788\n\n作用域含 runtime/test 的三份（runtime-acceptance / mainline-acceptance / revocation-sla）\n加上连带重写的 conformance-differential，在 HEAD 的干净检出 + 隔离栈 enterprise-platform-ms23\n重跑，全部 worktreeDirty=false：\n\n- runtime-acceptance   passed（7 步全绿）\n- conformance-differential passed\n- mainline-acceptance  passed（fact-read-http 11 / fact-persistence 17 / audit-tamper 19 /\n                                mainline-revocation-chaos 18）\n- revocation-sla       partial（p95 5036 ms，20 样本）\n\n结果与修复前一致，说明这次动的只是写报告的时机，没有动判据。\nui-acceptance / identity-ui 作用域不含 runtime/test，未受影响，保持原绑定。\n\nCo-Authored-By: Claude Opus 5 \u003cnoreply@anthropic.com\u003e\n","AuthorEmail":"hillao@juhailaoluodeMacBook-Pro.local","AuthorName":"juhailaoluo pro","CommitterEmail":"hillao@juhailaoluodeMacBook-Pro.local","CommitterName":"juhailaoluo pro","Timestamp":"2026-09-18T07:11:12-07:00"},{"Sha1":"925eee8195821074b0e2a699acc87b4af2ea35c4","Message":"docs(部署,runbook): 记录补 §10——:68 已修、两个方向的负向实测与隔离栈会被清空的教训\n\n- runbook §7 第四处那段的收尾从「代码修复尚未做」改成已落 08c3788，并写清改的是调用点不是\n  删掉占位写；那句「跑之前先确认这份报告已提交」现在只对「跑到一半失败」成立\n- 部署记录补 §10：负向实测两个方向（旧代码毁报告 / 新代码 sha256 不变）、正向确认 t+85s\n  仍写 RUN_NOT_COMPLETED、四份重跑结果，以及中途踩的一脚——enterprise-platform-ms23 是\n  具名共享验收栈且无卷，库和角色会被别的会话清掉，每轮开跑前重新供给比事后排查便宜\n\nCo-Authored-By: Claude Opus 5 \u003cnoreply@anthropic.com\u003e\n","AuthorEmail":"hillao@juhailaoluodeMacBook-Pro.local","AuthorName":"juhailaoluo pro","CommitterEmail":"hillao@juhailaoluodeMacBook-Pro.local","CommitterName":"juhailaoluo pro","Timestamp":"2026-09-18T07:11:12-07:00"},{"Sha1":"08c3788da42c5e8a4949daa8d5c7f23d30016205","Message":"fix(runtime): revocation-sla 的「已开跑」标记挪到前置校验之后——先毁证据再校验是反的\n\nadda76a 把三个验收 runner 改成「前置失败即退出、不碰报告」，但 revocation-sla 的报告不是\nrunner 写的，是用例自己写的，所以那次没覆盖到它——这是同病的第四处（runbook §7，2026-09-18 补注）。\n\n原形态：beforeAll 的**第一句**就无条件写 status=failed / RUN_NOT_COMPLETED，**下一句**才做\nisolated() 白名单校验。于是「环境没配对、这一轮根本没开始」也会先把上一轮的好报告毁掉。\n2026-09-18 重跑六份验收时第一次就踩到：模块库用了不在白名单里的名字，绑 765300b 的 partial\n当场没了（只因为跑在独立检出里才幸存）。\n\n改法不是删掉这个标记——「跑起来但没跑完」确实该在盘上留 RUN_NOT_COMPLETED 而不是上一轮的\npartial，这是用例自己写报告带来的责任。改的是**调用点**：挪到全部前置都满足之后，即模块库白名单\n通过、真实连库建好动作、API 子进程起来且 /health 就绪的那一刻。\n\n负向实测（两个方向都做了，确认这条改动真的有牙）：\n- 同一条件下跑旧代码：报告被改写成 RUN_NOT_COMPLETED（sha256 变化，好报告已毁）\n- 同一条件下跑新代码：测试照样以 MAINLINE_ISOLATED_DATABASE_REQUIRED 失败，报告 sha256 逐字节不变\n两次都用 DATABASE_URL_PERMISSION=…/platform_permission_notallowed 触发，5 例 skipped。\n\ntypecheck 通过（platform-tests 无 lint 配置，只有 test / typecheck 两个脚本）。\n受影响报告（runtime-acceptance / mainline-acceptance / revocation-sla，作用域含 runtime/test）\n随后在干净检出重跑回绑。\n\nCo-Authored-By: Claude Opus 5 \u003cnoreply@anthropic.com\u003e\n","AuthorEmail":"hillao@juhailaoluodeMacBook-Pro.local","AuthorName":"juhailaoluo pro","CommitterEmail":"hillao@juhailaoluodeMacBook-Pro.local","CommitterName":"juhailaoluo pro","Timestamp":"2026-09-18T02:51:53-07:00"},{"Sha1":"1f6fa2521e4d947ecd0ee43b18e691f836522049","Message":"docs(部署): 记录补 §9——六份验收在隔离栈重跑的结果与三个未写全的前置\n\n六份全部真跑回来（4 passed / 1 partial 设计内 / 1 passed），check:evidence 36 新鲜 / 4 过期。\n补上 runbook §7 没写全的三条：模块库后缀白名单、必须非超级用户角色、模块迁移要先 deploy。\n并登记 revocation-sla.test.ts:68 先毁证据再校验前置的缺陷与不在本轮修的理由。\n\nCo-Authored-By: Claude Opus 5 \u003cnoreply@anthropic.com\u003e\n","AuthorEmail":"hillao@juhailaoluodeMacBook-Pro.local","AuthorName":"juhailaoluo pro","CommitterEmail":"hillao@juhailaoluodeMacBook-Pro.local","CommitterName":"juhailaoluo pro","Timestamp":"2026-09-18T02:47:01-07:00"}],"HeadCommit":{"Sha1":"05e3826e9e9e4dbbdd753252c2134b8494b08f2b","Message":"chore(reports): 工作台两份快照回绑 @ dfd4832\n\ndocs/runbook.md 在 workbench-ops-snapshot 的作用域里且是 error 级——改 §7 那段就会当场\n把 check:evidence 打红。经 reports:rebind 在 HEAD 的干净检出里重跑，两份同源快照一并带回。\n\nCo-Authored-By: Claude Opus 5 \u003cnoreply@anthropic.com\u003e\n","AuthorEmail":"hillao@juhailaoluodeMacBook-Pro.local","AuthorName":"juhailaoluo pro","CommitterEmail":"hillao@juhailaoluodeMacBook-Pro.local","CommitterName":"juhailaoluo pro","Timestamp":"2026-09-18T07:11:36-07:00"},"CompareURL":"luoanwu/enterprise-platform/compare/1b2328540134f0118d2e3ed9abe10095e2ae4679...05e3826e9e9e4dbbdd753252c2134b8494b08f2b","Len":6}

TEXT

created_unix

INTEGER

Update row 31140 in action