Invariants / DC-NODE-26

DC-NODE-26

DC derived enforced

Decision / durable reconciliation (rung-2). After any applied receive decision, the chain_selector orchestrator's selector.current_tip EQUALS the durable ChainDb::tip (and the orchestrator chain_dep equals the durable PraosChainDepState). The in-memory decision state never diverges from the persisted authority: the orchestrator decides, the durable lockstep path applies, and the two are reconciled every decision. No applied decision leaves the selector ahead of, behind, or forked from the durable spine.

Source

docs/planning/phase4-n-ai-live-fork-choice-invariants.md (I-7)

Introduced in
PHASE4-N-AI

Enforcement trace

Tests 2

  • apply_reconciliation_mismatch_fails_fast
  • apply_rejected_makes_no_durable_change

Cross-references

Strengthened in