07 — Full Dry-Run Result (Supertrack G)

Full dry-run over all 39 PASS-eligible REQUIRED collections. No birth write. entity_code formula mirrors fn_birth_resolve_identity exactly and was cross-checked against fn_birth_register(...,dry_run=true) on real rows.

Summary (v_birth_qt001_dryrun_summary)

Metric	Value
Eligible collections	39
With non-zero delta	5
Empty (0 rows)	14
Fully covered (delta 0, rows>0)	20
Total source rows	7,195
Total existing births	7,466
Total expected new births	137

The 5 with non-zero delta

Collection	Strategy	src	existing	expected_delta
dot_domain_rules	synthetic_id	67	0	67
apr_approvals	synthetic_id	42	0	42
normative_relations	synthetic_id	18	0	18
apr_action_types	column:action_code	14	8	6
field_type_equivalences	synthetic_id	4	0	4

All 5 are governance_role='governed'. The 34 other eligible collections are already fully covered by realtime QT-002 (e.g. knowledge_documents 6586 rows / 6686 births / delta 0; task_comments 75/380/0) or empty.

Function-vs-formula cross-check

fn_birth_register('dot_domain_rules', row, true) → would_register / dot_domain_rules::1 (matches synthetic formula). checkpoint_instances rows → skipped (already_exists, delta 0). apr_request_types → blocked / identity_blocked_unclassified (correctly fail-closed). The view and the function agree.

No-go guard (v_birth_qt001_dryrun_no_go_guard) 4/4

no_negative_delta=0 · delta_never_exceeds_source=0 · zero_duplicate_risk_in_delta_scope=0 · eligible_count=39.

We know exactly what backfill would do: 137 governed births, byte-identical to what realtime would have produced, idempotent.