| content |
{"Commits":[{"Sha1":"feb7f6b07 {"Commits":[{"Sha1":"feb7f6b07069ba127bc49cce60d25f753d8a558e","Message":"K5.2b/K5.6: document the in-flight as-of window, registry count, ledger and candidate page\n\n- docs/metric-observation.md: recorded_at is insert time, not commit time; the\n repeatability claim is qualified for writes in flight (corrections,\n retractions and plain recordings alike). Lock order of RecordImpact.\n- docs/definition-registry.md: the Metric use count is historical observation\n rows; a correction adds one, a retracted observation still counts.\n- plan-ledger / gap-governance: K5.2b and K5.6 no longer say corrections,\n Impact state and independent review are not done; both stay partial.\n- docs/releases/candidates/k52b-k56-revision-review.md.\n\nCo-Authored-By: Claude Fable 5.1 \u003cnoreply@anthropic.com\u003e\n","AuthorEmail":"hillao@juhailaoluodeMacBook-Pro.local","AuthorName":"juhailaoluo pro","CommitterEmail":"hillao@juhailaoluodeMacBook-Pro.local","CommitterName":"juhailaoluo pro","Timestamp":"2026-10-03T00:44:18-07:00"},{"Sha1":"2ab4b38c9baab6cdec44a9e996495a87f83935d4","Message":"K5.2b/K5.6: hold the basis before judging it (Confirm, Declare, RecordImpact)\n\nConfirmAttribution, DeclareAttribution and RecordImpact judged their basis\nunder READ COMMITTED with only their own aggregate lock, so an in-flight\nRetractMetricObservation / correction, InvalidateImpact or InvalidateOutcome\nthat committed in between was missed. They now take, before the guards read\nand always in this order: the Impact advisory lock, the advisory lock of every\nmetric observation involved (ascending id), then a share lock on the Outcome\nrows (ascending id). The revising commands each hold exactly one of these.\n\ntest/value-basis-race.integration.test.js runs real parallel transactions,\nholding the first just before COMMIT and waiting until the database reports\nthe second blocked by it; 7 of its 9 tests fail on the previous source.\n\nCo-Authored-By: Claude Fable 5.1 \u003cnoreply@anthropic.com\u003e\n","AuthorEmail":"hillao@juhailaoluodeMacBook-Pro.local","AuthorName":"juhailaoluo pro","CommitterEmail":"hillao@juhailaoluodeMacBook-Pro.local","CommitterName":"juhailaoluo pro","Timestamp":"2026-10-03T00:44:18-07:00"},{"Sha1":"50da01cf4af1f54a79f9bf937ec3c37ae21bf09f","Message":"Merge origin/main (0.219.0 token hardening) into claude/gap2-observation-attribution-fixes\n\nKeeps both sides: 271 commands, 328 event types, 247 supporting events,\n92 append-only facts, 62 guarded child tables. Generated files re-rendered.\n\nCo-Authored-By: Claude Fable 5.1 \u003cnoreply@anthropic.com\u003e\n","AuthorEmail":"hillao@juhailaoluodeMacBook-Pro.local","AuthorName":"juhailaoluo pro","CommitterEmail":"hillao@juhailaoluodeMacBook-Pro.local","CommitterName":"juhailaoluo pro","Timestamp":"2026-10-03T00:23:24-07:00"}],"HeadCommit":{"Sha1":"feb7f6b07069ba127bc49cce60d25f753d8a558e","Message":"K5.2b/K5.6: document the in-flight as-of window, registry count, ledger and candidate page\n\n- docs/metric-observation.md: recorded_at is insert time, not commit time; the\n repeatability claim is qualified for writes in flight (corrections,\n retractions and plain recordings alike). Lock order of RecordImpact.\n- docs/definition-registry.md: the Metric use count is historical observation\n rows; a correction adds one, a retracted observation still counts.\n- plan-ledger / gap-governance: K5.2b and K5.6 no longer say corrections,\n Impact state and independent review are not done; both stay partial.\n- docs/releases/candidates/k52b-k56-revision-review.md.\n\nCo-Authored-By: Claude Fable 5.1 \u003cnoreply@anthropic.com\u003e\n","AuthorEmail":"hillao@juhailaoluodeMacBook-Pro.local","AuthorName":"juhailaoluo pro","CommitterEmail":"hillao@juhailaoluodeMacBook-Pro.local","CommitterName":"juhailaoluo pro","Timestamp":"2026-10-03T00:44:18-07:00"},"CompareURL":"luoanwu/seven-domain-five-dimension-unified-model/compare/159fb0eb8d9c7aaf40aa381bee5a3e527d02ae00...feb7f6b07069ba127bc49cce60d25f753d8a558e","Len":3}... |