There are a number of bugs and unimplemented features that make this code unsuitable for production use. Seek out other tools.
Nova gives your L2 contracts the power to read and write to L1 with minimal latency and no trust tradeoffs.
L1_NovaExecutionManager.sol
: Entry point for relayers to execute requests.L2_NovaRegistry.sol
: Hub for contracts/users on L2 to create and manage requests.L1_NovaApprovalEscrow.sol
: Escrow contract for relayers to approve input tokens to.libraries/
: Utilities used to help implement the Nova protocol.SafeTransferLib.sol
: Library for safely transferring Ether.NovaExecHashLib.sol
: Library for computing a Nova execHash.
external/
: Contracts and interfaces modified from external codebases.CrossDomainEnabled.sol
: Cross-domain communication helper mixin.
Below is a list of scripts used to test, fuzz, and measure the gas consumption of the Nova smart contracts. Many of these scripts are run automatically as part of our continuous integration suite.
To fail tests when their gas snapshots are incorrect (default is only a warning), set the CI
env var to true
.
npm run unit-tests
You must start up an instance of Optimism's "ops" repo before running integration tests.
npm run integration-tests
If you make a contribution that changes the gas usage of the contracts, you must run this command before committing.
npm run gas-changed
After running tests with coverage, an lcov report will be exported to coverage/index.html
.
npm run coverage
You must set the PRIVATE_KEY
and KOVAN_RPC_URL
environment variables before running integration tests on Kovan.
npm run kovan-integration-tests
You must install Echidna before fuzzing.
npm run fuzz deep {{CONTRACT_NAME}}
Replace {{CONTRACT_NAME}}
with a contract that is fuzzed in contracts/echidna
. A full list can be found here:
https://github.com/Rari-Capital/nova/blob/master/.github/workflows/fuzz.yml#L13-L14
There are 3 fuzz "modes" setup for this project:
-
deep
enters coverage guided inputs until it is halted manually.- It uses coverage guided fuzzing, which makes it quite slow.
- It is the most comprehensive mode (if run for long enough).
-
long
enters random inputs for 5 hours before halting.- It does not use coverage guided fuzzing.
- Is less comprehensive than
deep
.
-
quick
enters random inputs for 20 minutes before halting.- It does not use coverage guided fuzzing.
- Is less comprehensive than
long
.
To use any of these modes simply run the command above but replace deep
with the mode you wish to use (long
,quick
, or deep
).