12000x · 08 — Regression matrix (post-proofs vs 11000x baseline)

Every row re-verified live AFTER the four proof phases ran and rolled back. Δ = difference from 11000x baseline.

Repo + tests

Surface	11000x	12000x	Δ
Commit	3d56afa	3d56afa	none
Branch	main	main	none
Tree	clean	clean	none
pytest full suite	1324 PASS	1324 PASS	none

(12000x is pure proof — no code authored — so test count is expected to hold.)

Postgres (db=directus, schema=public)

Object	11000x	12000x	Δ
information_unit rows	163	163	none
iu_piece_collection rows	3	3	none
iu_piece_membership rows	8	8	none
iu_lifecycle_log rows	146	146	none
event_outbox rows total	122308	122308	none
event_outbox where event_domain='piece'	0	0	none
iu_vector_sync_point rows	152	152	none
Vector boundary violations	0	0	none
event_type_registry event_domain='piece' active	6	6	none

Qdrant

Collection	11000x	12000x	Δ
iu_core_iu_chunks points_count	149	149	none
production_documents points_count	9423 (never ours)	9423	none

Gate state (dot_config rows)

Key	11000x	12000x	Δ	Required?
iu_core.composer_enabled	false	false	none	must stay false
iu_core.routes_master_enabled	true	true	none	leave as-is
iu_core.route_worker_enabled	true	true	none	leave as-is
iu_core.delivery_live_routes	(empty)	(empty)	none	leave as-is
iu_core.retention_enabled	false	false	none	must stay false
iu_core.vector_sync_enabled	false	false	none	must stay false
iu_core.structure_ops_enabled	false	false	none	must stay false
piece_event_runtime.emit_enabled	false	false	none	must stay false
piece_event_runtime.dry_run_only	true	true	none	must stay true

All gates back to inert defaults after BEGIN/ROLLBACK in Phases D-G.

Healthcheck surfaces (Mac cron */10 last fire)

overall_ok : True
8/8 surfaces ok=True:
  three_axis_cache       in_sync table=163 view=163
  directus_collection    163 rows / 1 read-permission
  qdrant_collection      iu_core_iu_chunks (149 indexed)
  auto_refresh_trigger   gate=false fires_24h=0
  vector_boundary        149 pts / 141 unique
  write_gates            all 6 inert
  operator_runtime       open_runs=0 failed_24h=0 active_leases=0
  piece_event_runtime    emit_enabled=false dry_run_only=true types=6

11000x headline ("8th healthcheck surface piece_event_runtime live ok=true") still holds.

External state

Surface	11000x	12000x	Δ
PR #669 (Huyen1974/web-test)	OPEN/MERGEABLE not merged	OPEN/MERGEABLE not merged	none
Nuxt deploy	not deployed	not deployed	none
Retention sweep	disabled	disabled	none
Mac cron */10 exit	0	0	none

Migration cadence

Migration	11000x	12000x	Note
029 piece_event_runtime_substrate	LIVE	LIVE	unchanged
030 event_outbox event_domain += 'piece'	LIVE	LIVE	unchanged + live-INSERT round-trip re-verified (G.F)
(no migration 031)	—	—	deliberately authored no new migration

Conclusion

Every regression-watched surface is at 11000x baseline. The macro changed ZERO durable state. 1324/1324 tests remain GREEN. The 12000x outcome is purely additive in evidence, not in state.