Skip to main content

Read and use the parity vectors

Status: ProvisionalStatus: Built, runs on simulatorStatus: Built, runs on simulator

Parity vectors describe an ordered protocol scenario and the exact outcomes a harness must compare after each checkpoint.

Read the data before the narrative​

Stage 1 generates vector files from its reference scenarios. The builder executes actions and records snapshots or expected refusals. Integers are decimal strings, actors are names that each harness maps to wallets, and t is the block time set before an action. Time does not decrease within a vector.

An expect checkpoint compares the fields it contains exactly. Those fields can include balances, auction state, the payout vault, treasury, and each holder’s pending amounts. Status is derived at the checkpoint’s time. An expect_revert checkpoint requires the expected refusal and unchanged state.

The scenario description helps explain intent, but the actions, checkpoints, and implementation decide what a replay actually proves. Review all three when a prose summary and observed outcome disagree.

These are synthetic inputs​

Vector files deliberately expose test signing keys, salts, income witnesses, and terms commitment openings. They are test material, not examples of information that belongs in a production report log. Wallet-bound commitments must be recomputed for the harness’s actual actor wallets.

A prefix is a limited result​

Stage 2’s integration report separates complete replay, replay with waivers, replay with known divergence, prefix replay, and vectors not yet replayed. A prefix proves only the actions reached before its stopping point. A passing reference replay also does not prove that an adapter executed the same operations against deployed programs or contracts.

When reviewing a result, retain the source pins, backend, chain personality, ending checkpoint, waived assertions, and any divergence. Compare those with the proposed feature before using the result to support a readiness claim.

What can go wrong​

A green label can conceal a narrower test scope than the reader expects. Test keys and witness openings also become harmful if copied into an environment containing real funds or personal evidence. Keep the fixtures synthetic and describe exactly what was compared.