Testing

Parano1d uses three layers of tests: crate-level invariants, cross-crate release tests and fresh-process live scenarios.

Pruv contracts

Run the compiler, CLI and native contract tests in release mode:

cargo test --release --locked -p noid_pruv -p noid_tx
PRUV_TEST_Z3=/path/to/z3 cargo test --release --locked -p noid_pruv -p noid_tx -- --include-ignored

The second command includes the explicit Z3 tests. They cover universal postconditions, counterexamples and correspondence between the SMT model and native execution for all opcodes, predicate sources, inversions and destinations. Source/native differential tests cover u64 boundaries, every context operand, instruction order, nested conditions and scratch reuse. Parser and CLI tests cover resource limits, duplicate bindings and output preservation.

Two additional live scripts use a loopback-only network namespace:

The first requires a stopped, passed isolated fixture with branch-a and branch-b directories, such as the boundary/reorg qualification fixture. The second uses the stopped output of the first. Both copy their inputs into fresh directories. Build isolated_v2_node with isolated-v2-fork-testnet and a matching isolated proof bank. Use CLI and miner executables from the same source version.

NOID_PRUV_LIVE_DIR=/absolute/path/to/fresh-output \
NOID_PRUV_SOURCE_FIXTURE=/absolute/path/to/passed-v2-fixture \
NOID_PRUV_NODE=/absolute/path/to/isolated_v2_node \
NOID_PRUV_CLI=/absolute/path/to/parano1d-cli \
NOID_PRUV_MINER=/absolute/path/to/parano1d-miner \
NOID_PRUV_BIN=/absolute/path/to/pruv \
NOID_PRUV_Z3=/absolute/path/to/z3 \
unshare --user --map-root-user --net python3 scripts/live_pruv_contract_scenario.py

For the CLI scenario, set NOID_PRUV_CLI_SOURCE to that passed, stopped output, choose a new NOID_PRUV_LIVE_DIR, keep the executable paths and invoke scripts/live_pruv_cli_scenario.py in the same kind of namespace. Each run saves report.json, logs, source and binary hashes, definitions and receipts.

Fast checks

Run before a focused change:

cargo fmt --all -- --check
cargo check --release --locked --workspace --all-targets
cargo test --release --locked -p CHANGED_CRATE

For protocol code, include direct dependants. A transaction change commonly requires at least:

cargo test --release --locked \
  -p noid_tx \
  -p noid_chain \
  -p noid_mempool \
  -p noid_miner \
  -p noid_rpc \
  -p noid_node

Proof kernels

Production kernel tests should run in release mode:

cargo test --locked --release \
  -p noid_core \
  -p noid_poseidon2b \
  -p noid-ivc-core

On x86-64, force the production floor:

NOID_CPU_BACKEND=pclmul \
  cargo test --locked --release \
  -p noid_core -p noid_poseidon2b -p noid-ivc-core

The scalar backend is used only for differential checking:

NOID_CPU_BACKEND=scalar \
  cargo test --locked --release \
  -p noid_core -p noid_poseidon2b

Release gates

scripts/build_release.sh authenticates and embeds the release matrix packs and smoke-tests the native executables. Run source checks and tests separately before packaging. The package smoke tests cover:

• hardware preflight;
• node help and startup boundary;
• CLI;
• external miner;
• packaged GUI self-check.

Do not treat a debug binary or pack-free development node as a block-production test.

Live scenarios

Live scripts create fresh data directories under target/live-tests. They exercise real processes, RPC, P2P, MDBX and the production proof path.

Each script documents its environment knobs and binary expectations at the top. Run it from the repository root:

python3 scripts/live_two_miner_fork_reorg_scenario.py

Boundary changes

Changes near finality, epochs or State expansion require explicit boundary vectors, not only a long-running happy path. Cover:

• heights 143, 144 and 145 for transaction anchors;
• 17- and 18-block fork depths;
• 18- and 19-block synchronization gaps;
• finalized expansion windows with 9/9 and 10/8 occupancy;
• restart with the complete 36-header expansion lookback;
• payout boundaries and the final allocation height;
• State expansion from one log_slots tier to the next;
• shared-path terminal round trips, malformed encodings and decode caps;
• mempool count/byte pressure at 50% and 80%, empty-pool reset and Auto retries;
• partial mempool recovery, duplicate deliveries and disconnected sources;
• Small/Large selection and reserved system-mint slots, including all-reserved fallback.

Tests should use isolated data directories. They may use different ports to run concurrently, but must not mutate fixtures or production data.

Live test scenarios

The live_v2_* scripts use isolated, loopback-only networks and binaries with early test heights. Read each script's required matrix bank, environment and stopped-fixture inputs before running it. Mainnet source uses H210537; a pack for an early test height cannot be used in a mainnet build.

Boundary vectors also cover H−1/H/H+1, both legacy predecessor classes, mixed ASERT intervals, annual issuance steps, contract deadlines, checked u64 arithmetic and competing spends. Full workloads share the page, input and call budgets. Validate the mainnet bank separately from the early-height bank; performance identifies measured artifacts.