Hardhat Plugin For Replicable Deployments And Tests
npm install @layerzerolabs/hardhat-deploy_A Hardhat Plugin For Replicable Deployments And Easy Testing_
> A complete dev template using hardhat-deploy is available here: https://github.com/wighawag/template-ethereum-contracts
> It also contains various branches exemplifying the capability of hardhat-deploy. Check it out.
- What is it for?
- hardhat-deploy in a nutshell
- Installation
- npm install hardhat-deploy
- TypeScript support
- Migrating existing deployment to hardhat-deploy
- Hardhat Tasks Available/Updated
- 1. hardhat deploy
- Options
- Flags
- 2. hardhat node
- Options
- Flags
- 3. hardhat test
- 4. hardhat etherscan-verify
- Options
- 5. hardhat sourcify
- Options
- 6. hardhat export
- Options
- Hardhat Environment Extensions
- Configuration
- 1. namedAccounts (ability to name addresses)
- 2. extra hardhat.config networks' options
- live
- saveDeployments
- tags
- deploy
- companionNetworks
- 3. extra hardhat.config paths' options
- 4. deterministicDeployment (ability to specify a deployment factory)
- Importing deployment from other projects (with truffle support)
- Access to Artifacts (non-deployed contract code and abi)
- How to Deploy Contracts
- The deploy Task
- Deploy Scripts
- An example of a deploy script :
- The deployments field
- deployments.deploy(
- Handling contract using libraries
- Exporting Deployments
- Deploying and Upgrading Proxies
- When the constructor and init functions are different
- Proxy deployment options
- Builtin-In Support For Diamonds (EIP2535)
- deployment / upgrade
- onUpgrade calls
- more...
- Testing Deployed Contracts
- Creating Fixtures
- More Information On Hardhat Tasks
- 1. node task
- 2. test task
- 3. run task
- 4. console task
- Deploy Scripts: Tags And Dependencies
This hardhat plugin adds a mechanism to deploy contracts to any network, keeping track of them and replicating the same environment for testing.
It also adds a mechanism to associate names to addresses, so test and deployment scripts can be reconfigured by simply changing the address a name points to, allowing different configurations per network. This also results in much clearer tests and deployment scripts (no more accounts[0] in your code).
This plugin contains a lot more features too, all geared toward a better developer experience :
- chain configuration export
- listing deployed contracts' addresses and their abis (useful for web apps)
- library linking at the time of deployment.
- deterministic deployment across networks.
- support for specific deploy script per network (L1 vs L2 for example)
- ability to access deployment from "companion" networks
- deployment dependency system (allowing you to only deploy what is needed).
- deployment retrying (by saving pending tx): so you can feel confident when making a deployment that you can always recover.
- deployments as test fixture using evm_snapshot to speed up testing.
- ability to create your own test fixture that automatically benefits from evm_snapshot's tests speed-up boost
- combined with hardhat-deploy-ethers it has the ability to get ethers contract instance by name (like await ethers.getContract("ContractName")).
- importing artifacts from external sources (like npm packages), including truffle support.
- importing deployments from external sources (like npm packages)
- ability to log information in deploy mode only (while in test the console remains clean).
- contains helpers to read and execute transaction on deployed contract referring to them by name.
- These helpers contains options to auto-mine on dev network (to speed up test deployments).
- save metadata of deployed contract so they can always be fully verified, via sourcify or etherscan.
- ability to submit contract source to etherscan and sourcify for verification at any time. (Because hardhat-deploy will save all the necessary info, it can be executed at any time.)
- support hardhat's fork feature so deployment can be accessed even when run through fork.
- named accounts are automatically impersonnated too, so you can perform tx as if you had their private key.
- proxy deployment with ability to upgrade them transparently, only if code changes.
- this include support for openzeppelin transparent proxies
- diamond deployment with facets, allowing you to focus on what the new version will be. It will generate the diamondCut necessary to reach the new state.
- watch and deploy: hardhat-deploy can watch both your deploy script and contract code and redeploy on changes.
- HCR (Hot Contract Replacement): the watch feature combined with proxy or diamond, gives you an experience akin to frontend Hot Module Replacement: once your contract changes, the deployment is executed and your contract retains the same address and same state, allowing you to tweak your contracts while debugging your front-end.
Before going into the details, here is a very simple summary of the basic feature of hardhat-deploy.
hardhat-deploy allows you to write deploy scripts in the deploy folder. Each of these files that look as follows will be executed in turn when you execute the following task: hardhat --network
``js`
// deploy/00_deploy_my_contract.js
module.exports = async ({getNamedAccounts, deployments}) => {
const {deploy} = deployments;
const {deployer} = await getNamedAccounts();
await deploy('MyContract', {
from: deployer,
args: ['Hello'],
log: true,
});
};
module.exports.tags = ['MyContract'];
Furthermore you can also ensure these scripts are executed in test too by calling await deployments.fixture(['MyContract']) in your test.
This is optimized, so if multiple tests use the same contract, the deployment will be executed once and each test will start with the exact same state.
This is a huge benefit for testing since you are not required to replicate the deployment procedure in your tests. The tag feature (as seen in the script above) and dependencies will also make your life easier when writing complex deployment procedures.
You can even group deploy scripts in different sub-folders and ensure they are executed in their logical order.
Furthermore hardhat-deploy can also support a multi-chain settings like L1, L2 with multiple deploy folder specific to each network.
All of this can also be bundled in a npm package so users of hardhat-deploy can reuse your deployment procedure and get started integrating with your project locally.
There is a tutorial covering the basics here: https://github.com/wighawag/tutorial-hardhat-deploy
`bash`
npm install -D hardhat-deploy
And add the following statement to your hardhat.config.js:
`js`
require('hardhat-deploy');
if you use ethers.js we recommend you also install hardhat-deploy-ethers which add extra features to access deployments as ethers contract.
`bash`
npm install --save-dev @nomiclabs/hardhat-ethers hardhat-deploy-ethers ethers
More details on hardhat-deploy-ethers repo: https://github.com/wighawag/hardhat-deploy-ethers#readme
With hardhat the tsconfig.json is optional.
But if you add folders to the include field in tsconfig.json, you ll also need to include hardhat.config.ts like :
"include": ["./hardhat.config.ts", "./scripts", "./deploy", "./test"]
for deploy script (see below) you can write them this way to benefit from typing :
`ts
import {HardhatRuntimeEnvironment} from 'hardhat/types';
import {DeployFunction} from 'hardhat-deploy/types';
const func: DeployFunction = async function (hre: HardhatRuntimeEnvironment) {
// code here
};
export default func;
`
See a template that uses hardhat-deploy here: https://github.com/wighawag/template-ethereum-contracts
This repo has also some examples branch that exemplify specific features, like the forking testing here: https://github.com/wighawag/template-ethereum-contracts/tree/examples/fork-test
> Only needed for an existing project that already deployed contracts and has the deployment information available (at minimum, address and abi)
You might want to switch your current deployment process to use hardhat-deploy. In that case you probably have some deployments saved elsewhere.
In order to port them to hardhat-deploy, you'll need to create one .json file per contract in the deployments/ folder (configurable via paths config).
The network folder is simply the hardhat network name (as configured in hardhat.config.js) (accessible at runtime via hre.network.name)..chainId
Such folder need to have a file named containing the chainId as decimal.
For example for a network named "rinkeby" (for the corresponding network) the file deployments/rinkeby/.chainId would be
`
4
Note, prior to hardhat 0.6 the chainId was appended to the folder name (expect for some known network name). This has changed and upgrading to 0.6 will require you to change the folder name and add the '.chainId' file.
Each contract file must follow this type (as defined in types.ts) :
`ts`
export interface Deployment {
address: Address;
abi: ABI;
receipt?: Receipt;
transactionHash?: string;
history?: Deployment[];
numDeployments?: number;
implementation?: string;
args?: any[];
linkedData?: any;
solcInputHash?: string;
metadata?: string;
bytecode?: string;
deployedBytecode?: string;
libraries?: Libraries;
userdoc?: any;
devdoc?: any;
methodIdentifiers?: any;
diamondCut?: FacetCut[];
facets?: Facet[];
storageLayout?: any;
gasEstimates?: any;
}
As you can see, only abi and address are mandatory. But having the other fields allow more feature. For example, metadata and args allow you to benefit from contract code verification.
For Receipt, the following type is expected:
`ts`
export type Receipt = {
from: string;
transactionHash: string;
blockHash: string;
blockNumber: number;
transactionIndex: number;
cumulativeGasUsed: string;
gasUsed: string;
contractAddress?: string;
to?: string;
logs?: Log[];
events?: any[];
logsBloom?: string;
byzantium?: boolean;
status?: number;
confirmations?: number;
};
Here is an example:
Let's say you have:
- 2 Contract named Greeter and Registry deployed on rinkeby
- 1 contract named Greeter on mainnet
- 2 Contract named Greeter and Registry deployed on a network named rinkeby2
You would get the following folder structure:
`
deployments/
mainnet/
.chainId
Greeter.json
rinkeby/
.chainId
Greeter.json
Registry.json
rinkeby2/
.chainId
Greeter.json
Registry.json
The reason why hardhat-deploy save chainId in the .chainId file is both for
- safety: so that if you were to change the network name to point to a different chain, it would not attempt to read the wrong folder and assume that a contract has been deployed while it has not.
- ability to know the chainId without requiring to be connected to a node (and so not dependent on hardhat.config.js settings). Useful for export task.
---
hardhat deploy adds several tasks to hardhat. It also modifies existing one, adding new options and new behavior. All of these are described here:
---
---
This plugin adds the _deploy_ task to Hardhat.
This task will execute the scripts in the deploy folder and save the contract deployments to disk. These deployments are supposed to be saved for example in a git repository. This way they can be accessed later. But you are free to save them elsewhere and get them back via your mechanism of choice.
With the deployment saved, it allows you to deploy a contract only if changes were made.
Deploy scripts (also called Deploy functions) can also perform arbitrary logic.
For further details on how to use it and write deploy script, see section below.
#### Options
--export : export one file that contains all contracts (address, abi + extra data) for the network being invoked. The file contains the minimal information so to not bloat your front end.
--export-all : export one file that contains all contracts across all saved deployments, regardless of the network being invoked.
--tags : only execute deploy scripts with the given tags (separated by commas) and their dependencies (see more info here about tags and dependencies)
--gasprice : specify the gasprice (in wei) to use by default for transactions executed via hardhat-deploy helpers in deploy scripts
--write : default to true (except for hardhat network). If true, write deployments to disk (in deployments path, see path config).
#### Flags
--reset: This flag resets the deployments from scratch. Previously deployed contracts are not considered and deleted from disk.
--silent: This flag removes hardhat-deploy log output (see log function and log options for hre.deployments)
--watch: This flag make the task never-ending, watching for file changes in the deploy scripts folder and the contract source folder. If any changes happen the contracts are recompiled and the deploy script are re-run. Combined with a proxy deployment (Proxies or Diamond) this allow to have HCR (Hot Contract Replacement).
---
---
This plugin modifies the _node_ task so that it also executes the deployment script before exposing the server http RPC interface
It adds similar options than the deploy task :
#### Options
--export : export one file that contains all contracts (address, abi + extra data) for the network being invoked. The file contains the minimal information so to not bloat your front end. If the extension ends in .ts it will generate a typescript file containing the contracts info.
--export-all : export one file that contains all contracts across all saved deployment, regardless of the network being invoked. If the extension ends in .ts it will generate a typescript file containing the contracts info.
--tags : only excutes deploy scripts with the given tags (separated by commas) and their dependencies (see more info here about tags and dependencies)
--gasprice : specify the gasprice to use by default for transactions executed via hardhat-deploy helpers in deploy scripts
--write : default to true (except for hardhat network). If true, write deployments to disk (in deployments path, see path config).
#### Flags
--no-reset: This flag prevents the resetting of the existing deployments. This is usually not desired when running the node task as a network is created from scratch and previous deployments are irrelevant.
--silent: This flag removes hardhat-deploy log output (see log function and log options for hre.deployments)
--silent: This flag removes hardhat-deploy log output (see log function and log options for hre.deployments)
--watch: This flag makes the task never-ending, watching for file changes in the deploy scripts folder and the contract source folder. If any changes happen the contracts are recompiled and the deploy script are re-run. Combined with a proxy deployment (Proxies or Diamond) this allows to have HCR (Hot Contract Replacement).
--no-deploy that discard all other options to revert to normal hardhat node behavior without any deployment being performed.
> :warning: Note that the deployments are saved as if the network name is localhost. This is because hardhat node is expected to be used as localhost: You can for example execute hardhat --network localhost console after node is running. Doing hardhat --network hardhat console would indeed not do anything useful. It still takes the configuration from hardhat in the hardhat.config.js file though.
---
---
This plugin adds a flag argument --deploy-fixture to the _test_ task which if enabled will run the global deployments fixture before the tests and snapshots it. This will generally speed up the tests as further test will be able to revert back to the full deployment.
> :warning: Note though that if your test behave differently whether that option is on or not, this most likely means that your deploy scripts' tags and dependencies are not configured correctly. This is because the global fixture will ensure all contracts are deployed while test will usually (for efficiency) ask for a particular tag.
---
---
This plugin adds the _etherscan-verify_ task to Hardhat.
This task will submit the contract source and other info of all deployed contracts to allow etherscan to verify and record the sources.
Instead of using the full solc input, this task will first attempt to send the minimal sources from the metadata.
But Etherscan sometime fails due to a bug in solidity compiler (
This task will also attempt to automatically find the SPDX license in the source.
To execute that task, you need to specify the network to run against :
`bash`
hardhat --network mainnet etherscan-verify [--api-key
#### Options
Note that hardhat-deploy now use a different config format to not conflict with hardhat-etherscan
--api-key : let you specify your etherscan api key. Alternatively, you can provide it via the env variable ETHERSCAN_API_KEY or through the hardhat.config.ts verify field:
`js`
{
...
verify: {
etherscan: {
apiKey: '
}
}
}
Keep in mind that the ETHERSCAN_API_KEY .env variable is read first, before the hardhat.config.ts. ETHERSCAN_API_KEY`
If you want to set up multi-network api key support, you can do it by adding an env loader that use .env. for each network this way. ETHERSCAN_API_KEY
Alternatively, you can change the mainnet etherscan api key env var name to something other than , and specify the other network keys as specified above.
--api-url : let you specify your etherscan url to submit the source to. Can also be configured per network in hardhat.config.js:
`js`
{
...
networks: {
mynetwork: {
...
verify: {
etherscan: {
apiUrl: 'http://mynetwork.xyz'
}
}
}
}
}
--license : SPDX license (useful if SPDX is not listed in the sources), need to be supported by etherscan: https://etherscan.io/contract-license-types
--force-license: if set, will force the use of the license specified by --license option, ignoring the one in the source (useful for license not supported by etherscan)
--solc-input: fallback on solc-input id needed (useful when etherscan fails on the minimum sources, see https://github.com/ethereum/solidity/issues/9573)
--sleep: sleep 500ms between each verification, so API rate limit is not exceeded
---
---
This plugin adds the _sourcify_ task to Hardhat.
Similar to hardhat etherscan-verify this task will submit the contract source and other info of all deployed contracts to sourcify.
`bash`
hardhat --network mainnet sourcify
Later this task might instead pin the metadata to ipfs, so sourcify can automatically verify them.
#### Options
--contract-name : specify the contract's name you want to verify
--endpoint : specify the sourcify endpoint, default to https://sourcify.dev/server/
--write-failing-metadata: if set and the sourcify task fails to verify, the metadata file will be written to disk, so you can more easily figure out what has gone wrong.
---
---
This plugin adds the _export_ task to Hardhat.
This task will export the contract deployed (saved in deployments folder) to a file with a simple format containing only contract addresses and abi, useful for web apps.
One of the following options need to be set for this task to have any effects :
#### Options
--export : export one file that contains all contracts (address, abi + extra data) for the network being invoked. The file contains the minimal information so to not bloat your front end. If the extension ends in .ts it will generate a typescript file containing the contracts info.
--export-all : export one file that contains all contracts across all saved deployment, regardless of the network being invoked. If the extension ends in .ts it will generate a typescript file containing the contracts info.
This last option has some limitations, when combined with the use of external deployments (see Configuration). If such external deployments were using older version of hardhat-deploy or truffle, the chainId might be missing. In order for these to be exported, the hardhat network config need to explicity state the chainId in the networks config of hardhat.config.js.
With both --export and --export-all, using the special value of - will output to STDOUT rather than writing a normal file.
---
---
This plugin extends the Hardhat Runtime Environment by adding 4 fields:
- getNamedAccounts: () => Promise<{ [name: string]: string }>: a function returning an object whose keys are names and values are addresses. It is parsed from the namedAccounts configuration (see Configuration).
- getUnnamedAccounts: () => Promise: accounts which has no names, useful for test where you want to be sure that the account is not one of the predefined one
- deployments: contains functions to access past deployments or to save new ones, as well as helpers functions.
- getChainId(): Promise: offer an easy way to fetch the current chainId.
---
---
---
---
This plugin extends the HardhatConfig's object with an optional namedAccounts field.
namedAccounts allows you to associate names to addresses and have them configured per chain.
This allows you to have meaningful names in your tests while the addresses match to multi sig in real network for example.
`js`
{
namedAccounts: {
deployer: {
default: 0, // here this will by default take the first account as deployer
1: 0, // similarly on mainnet it will take the first account as deployer. Note though that depending on how hardhat network are configured, the account 0 on one network can be different than on another
4: '0xA296a3d5F026953e17F472B497eC29a5631FB51B', // but for rinkeby it will be a specific address
"goerli": '0x84b9514E013710b9dD0811c9Fe46b837a4A0d8E0', //it can also specify a specific netwotk name (specified in hardhat.config.js)
},
feeCollector:{
default: 1, // here this will by default take the second account as feeCollector (so in the test this will be a different account than the deployer)
1: '0xa5610E1f289DbDe94F3428A9df22E8B518f65751', // on the mainnet the feeCollector could be a multi sig
4: '0xa250ac77360d4e837a13628bC828a2aDf7BabfB3', // on rinkeby it could be another account
}
}
}
---
---
hardhat-deploy add 5 new fields to networks configuration
#### live
this is not used internally but is useful to perform action on a network whether it is a live network (rinkeby, mainnet, etc) or a temporary one (localhost, hardhat). The default is true (except for localhost and hardhat where the default is false).
#### saveDeployments
this tell whether hardhat-deploy should save the deployments to disk or not. Default to true, except for the hardhat network.
#### tags
network can have tags to represent them. The config is an array and at runtime the hre.network.tags is an object whose fields (the tags) are set to true.
This is useful to conditionally operate on network based on their use case.
Example:
`js`
{
networks: {
localhost: {
live: false,
saveDeployments: true,
tags: ["local"]
},
hardhat: {
live: false,
saveDeployments: true,
tags: ["test", "local"]
},
rinkeby: {
live: true,
saveDeployments: true,
tags: ["staging"]
}
}
}
#### deploy
the deploy field overrides the paths.deploy option and let you define a set of folder containing the deploy scripts to be executed for this network.
You can thus have one network that will be executing L1 deployment and other L2 deployments, etc...
You could also have a folder that deploy contracts that are live on mainnet but that you need to replicate for your test or local network.
Example:
`js`
{
networks: {
mainnet: {
deploy: [ 'deploy/' ]
},
rinkeby: {
deploy: [ 'testnet-deploy/' ]
}
}
}
#### companionNetworks
the companionNetworks field is an object whose key is any name you desire and the value is the name of a network that will be accessible inside the deploy script. For example:
`js`
{
...
networks: {
optimism: {
url: 'http://127.0.0.1:8545',
ovm: true,
companionNetworks: {
l1: 'localhost',
},
}
}
...
}
By using name you can have the same deploy script used in different set of network.
For your test you could have the companion networks pointing to the same hardhat network, for test deployment, you could have rinkeby acting like your l2 while goerli act as your l1.
An example repo that showcase a multi-network setup with optimism can be found here: https://github.com/wighawag/template-ethereum-contracts/tree/examples/optimism
deploy script can then access the network and its deployment as follow :
`js
module.exports = async ({
getNamedAccounts,
deployments,
getChainId,
getUnnamedAccounts,
}) => {
const {deploy, execute} = deployments;
const {deployer} = await getNamedAccounts();
const OVM_L1ERC20Gateway = await hre.companionNetworks['l1'].deployments.get(
'OVM_L1ERC20Gateway'
); // layer 1
await execute(
'SimpleERC20_OVM',
{from: deployer, log: true},
'init',
OVM_L1ERC20Gateway.address
);
};
`
---
---
hardhat-deploy also adds fields to HardhatConfig's ProjectPaths object.
Here is an example showing the default values :
`js`
{
paths: {
deploy: 'deploy',
deployments: 'deployments',
imports: 'imports'
}
}
The deploy folder is expected to contain the deploy scripts that are executed upon invocation of hardhat deploy or hardhat node.
It can also be an array of folder path.
The deployments folder will contain the resulting deployments (contract addresses along their abi, bytecode, metadata...). One folder per network and one file per contract.
The imports folder is expected to contain artifacts that were pre-compiled. Useful if you want to upgrade to a new solidity version but want to keep using previously compiled contracts. The artifact is the same format as normal hardhat artifact, so you can easily copy them over, before switching to a new compiler version.
> This is less useful now that hardhat support multiple solidity compiler at once.
---
---
---
This plugin extends the HardhatConfig's object with an optional deterministicDeployment field.
deterministicDeployment allows you to associate information that are used on each network for deterministic deployment. The information for each deterministic deployment consist out of a factory, a deployer, the required funding and a signedTx to deploy the factory. The default deterministic deployment used is the Deterministic Deployment Proxy. The factory expects a 32 bytes salt concatenated with the deployment data (see EIP-1014 for more information on these parameters).
Using the deterministicDeployment it is possible to define a different setup for the deterministic deployment. One use case for this is the deterministic deployment on networks that required replay protection (such as Celo or Avalanche). The Deterministic Deployment Proxy can only be deployed on networks that don't enforce replay protection, therefore on other networks an alternative library has to be used. An example for this would be the Safe Singleton Factory that is an adjusted version of the Deterministic Deployment Proxy that contains signed transactions that include replay protection.
The information can be defined either as an object
`js`
{
deterministicDeployment: {
"4": {
factory: "
deployer: "
funding: "
signedTx: "
}
}
}
or as a function that returns the information for the deterministic deployment
`js`
{
deterministicDeployment: (network: string) => {
return {
factory: '
deployer: '
funding: '
signedTx: '
};
};
}
hardhat-deploy also add the external field to HardhatConfig
Such field allows to specify paths for external artifacts or deployments.
The external object has 2 fields:
`js`
{
external: {
contracts: [
{
artifacts: "node_modules/@cartesi/arbitration/export/artifacts",
deploy: "node_modules/@cartesi/arbitration/export/deploy"
},
{
artifacts: "node_modules/someotherpackage/artifacts",
}
],
deployments: {
rinkeby: ["node_modules/@cartesi/arbitration/build/contracts"],
},
}
}
The contracts field specify an array of object which itself have 2 fields.
- artifacts: (mandatory) it is a path to an artifact folder. This support both hardhat and truffle artifacts.deploy
- : (optional) it specifies a path to a folder where reside deploy script. The deploy scripts have only access to the artifact specified in the artifacts field. This allow project to share their deployment procedure. A boon for developer aiming at integrating it as they can get the contracts to be deployed for testing locally.
The deployments fields specify an object whose field names are the hardhat network and the value is an array of path to look for deployments. It supports both hardhat-deploy and truffle formats.
---
Artifacts in hardhat terminology represent a compiled contract (not yet deployed) with at least its bytecode and abi.
hardhat-deploy gives can access to these artifacts via the deployments.getArtifact function :
`js`
const {deployments} = require('hardhat');
const artifact = await deployments.getArtifact(artifactName);
With the hardhat-deploy-ethers plugin you can get an artifact as an ethers contract factory, ready to be deployed, via the following:
`js`
const {deployments, ethers} = require('hardhat');
const factory = await ethers.getContractFactory(artifactName);
> Note that the artifact's files need to be either in artifacts folder that hardhat generate on compilation or in the imports folder where you can store contracts compiled elsewhere. They can also be present in the folder specified in external.artifacts see Importing deployment from other projects
---
---
---
hardhat --network
This is a new task that the hardhat-deploy adds. As the name suggests it deploys contracts.deploy
To be exact it will look for files in the folder or whatever was configured in paths.deploy, see paths config
It will scan for files in alphabetical order and execute them in turn.
- it will require each of these files and execute the exported function with the HRE as argument
To specify the network, you can use the builtin hardhat argument --network or set the env variable HARDHAT_NETWORK
> :warning: Note that running hardhat deploy without specifying a network will use the default network. If the default network is hardhat (the default's default) then nothing will happen as a result as everything happens in memory, but this can be used to ensure the deployment is without issues.
---
The deploy scripts need to be of the following type :
`ts`
export interface DeployFunction {
(env: HardhatRuntimeEnvironment): Promise
skip?: (env: HardhatRuntimeEnvironment) => Promise
tags?: string[];
dependencies?: string[];
runAtTheEnd?: boolean;
id?: string;
}
The skip function can be used to skip executing the script under whatever condition. It simply need to resolve a promise to true.
The tags is a list of string that when the _deploy_ task is executed with, the script will be executed (unless it skips). In other word if the deploy task is executed with a tag that does not belong to that script, that script will not be executed unless it is a dependency of a script that does get executed.
The dependencies is a list of tag that will be executed if that script is executed. So if the script is executed, every script whose tag match any of the dependencies will be executed first.
The runAtTheEnd is a boolean that if set to true, will queue that script to be executed after all other scripts are executed.
These set of fields allow more flexibility to organize the scripts. You are not limited to alphabetical order and you can even organise deploy script in sub folders.
Finally the function can return true if it wishes to never be executed again. This can be useful to emulate migration scripts that are meant to be executed only once. Once such script return true (async), the id field is used to track execution and if that field is not present when the script return true, it will fails.
In other words, if you want a particular deploy script to run only once, it needs to both return true (async) and have an id set.
In any case, as a general advice every deploy function should be idempotent. This is so they can always recover from failure or pending transaction. This is what underpin most of hardhat-deploy philosophy.
This is why the hre.deployments.deploy function will by default only deploy if the contract code has changed, making it easier to write idempotent script.
#### An example of a deploy script :
`js
module.exports = async ({
getNamedAccounts,
deployments,
getChainId,
getUnnamedAccounts,
}) => {
const {deploy} = deployments;
const {deployer} = await getNamedAccounts();
// the following will only deploy "GenericMetaTxProcessor" if the contract was never deployed or if the code changed since last deployment
await deploy('GenericMetaTxProcessor', {
from: deployer,
gasLimit: 4000000,
args: [],
});
};
`
As you can see the HRE passed in has 4 new fields :
- getNamedAccounts is a function that returns a promise to an object whose keys are names and values are addresses. It is parsed from the namedAccounts configuration (see namedAccounts).
- getUnnamedAccounts is a function that return a promise to an array of accounts which were not named (see namedAccounts). It is useful for tests where you want to be sure that the account has no speicifc role in the system (no token given, no admin access, etc...).
- getChainId is a function which return a promise for the chainId, as convenience
- deployments is an object which contains functions to access past deployments or to save new ones, as well as helpers functions.
That latter field contains for example the deploy function that allows you to deploy contract and save them. It contains a lot more functions though :
---
The deployments field contains several helpers function to deploy contract but also execute transaction.
`tssalt
export interface DeploymentsExtension {
deploy(name: string, options: DeployOptions): Promise
diamond: {
// deploy diamond based contract (see section below)
deploy(name: string, options: DiamondOptions): Promise
};
deterministic( // return the deterministic address as well as a function to deploy the contract, can pass the field in the option to use different salt
name: string,
options: Create2DeployOptions
): Promise<{
address: Address;
implementationAddress?: Address;
deploy(): Promise
}>;
fetchIfDifferent( // return true if new compiled code is different than deployed contract
name: string,
options: DeployOptions
): Promise<{differences: boolean; address?: string}>;
save(name: string, deployment: DeploymentSubmission): Promise
get(name: string): Promise
getOrNull(name: string): Promise
getDeploymentsFromAddress(address: string): Promise
all(): Promise<{[name: string]: Deployment}>; // return all deployments
getArtifact(name: string): Promise
getExtendedArtifact(name: string): Promise
run( // execute deployment scripts
tags?: string | string[],
options?: {
resetMemory?: boolean;
deletePreviousDeployments?: boolean;
writeDeploymentsToFiles?: boolean;
export?: string;
exportAll?: string;
}
): Promise<{[name: string]: Deployment}>;
fixture( // execute deployment as fixture for test // use evm_snapshot to revert back
tags?: string | string[],
options?: {fallbackToGlobal?: boolean; keepExistingDeployments?: boolean}
): Promise<{[name: string]: Deployment}>;
createFixture
func: FixtureFunc
id?: string
): (options?: O) => Promise
log(...args: any[]): void; // log data only ig log enabled (disabled in test fixture)
execute( // execute function call on contract
name: string,
options: TxOptions,
methodName: string,
...args: any[]
): Promise
rawTx(tx: SimpleTx): Promise
catchUnknownSigner( // you can wrap other function with this function and it will catch failure due to missing signer with the details of the tx to be executed
action: Promise
options?: {log?: boolean}
): Promise
to?: string;
value?: string;
data?: string;
}>;
read( // make a read-only call to a contract
name: string,
options: CallOptions,
methodName: string,
...args: any[]
): Promise
read(name: string, methodName: string, ...args: any[]): Promise
// rawCall(to: Address, data: string): Promise
}
`
---
#### deployments.deploy(
---
The deploy function, as mentioned above, allows you to deploy a contract and save it under a specific name.
The deploy function expect 2 parameters: one for the name and one for the options
See below the full list of fields that the option parameter allows and requires:
`tsgetNamedAccounts
export interface DeployOptions = {
from: string; // address (or private key) that will perform the transaction. you can use to retrieve the address you want by name.
contract?: // this is an optional field. If not specified it defaults to the contract with the same name as the first parameter
| string // this field can be either a string for the name of the contract
| { // or abi and bytecode
abi: ABI;
bytecode: string;
deployedBytecode?: string;
};
args?: any[]; // the list of argument for the constructor (or the upgrade function in case of proxy)
skipIfAlreadyDeployed?: boolean; // if set it to true, will not attempt to deploy even if the contract deployed under the same name is different
log?: boolean; // if true, it will log the result of the deployment (tx hash, address and gas used)
linkedData?: any; // This allow to associate any JSON data to the deployment. Useful for merkle tree data for example
libraries?: { [libraryName: string]: Address }; // This let you associate libraries to the deployed contract
proxy?: boolean | string | ProxyOptions; // This options allow to consider your contract as a proxy (see below for more details)
// here some common tx options :
gasLimit?: string | number | BigNumber;
gasPrice?: string | BigNumber;
value?: string | BigNumber;
nonce?: string | number | BigNumber;
estimatedGasLimit?: string | number | BigNumber; // to speed up the estimation, it is possible to provide an upper gasLimit
estimateGasExtra?: string | number | BigNumber; // this option allow you to add a gas buffer on top of the estimation
autoMine?: boolean; // this force a evm_mine to be executed. this is useful to speed deployment on test network that allow to specify a block delay (ganache for example). This option basically skip the delay by force mining.
deterministicDeployment? boolean | string; // if true, it will deploy the contract at a deterministic address based on bytecode and constructor arguments. The address will be the same across all network. It use create2 opcode for that, if it is a string, the string will be used as the salt.
waitConfirmations?: number; // number of the confirmations to wait after the transactions is included in the chain
};
`
---
---
In the deploy function, one of the DeployOptions field is the libraries field. It allows you to associate external contract as libraries at the time of deployment.
First, you have deploy the library using the deploy function, then when we deploy a contract that needs the linked library, we can pass the deployed library name and address in as an argument to the libraries object.
First step: deploy the library:
`js
const exampleLibrary = await deploy("ExampleLibary", {
from:
});
`
ExampleLibrary is now deployed to whatever network was chosen (hardhat deploy --network )
For example, if we are deploying on Rinkeby, this library will get deployed on rinkeby, and the exampleLibrary variable will be a deployment object that contains the abi as well as the deployed address for the contract.
Now that the library is deployed, we can link it in our next deployed contract.
`js
const example = await deploy("Example", {
from:
args: ["example string argument for the 'Example' contract constructor"],
libraries: {
ExampleLibrary: exampleLibrary.address
}
});
`
This libraries object takes the name of the library, and its deployed address on the network. Multiple libraries can be passed into the libraries object.
---
---
Apart from deployments saved in the deployments folder which contains all information available about the contract (compile time data + deployment data), hardhat-deploy allows you to export lightweight files.
These can be used for example to power your frontend with contract's address and abi.
This come into 2 flavors.
The first one is exported via the --export option and follow the following format :
`ts`
export interface Export {
chainId: string;
name: string;
contracts: {[name: string]: ContractExport};
}
where name is the name of the network configuration chosen (see hardhat option --network)
The second one is exported via the --export-all option and follow the following format :
`ts`
export type MultiExport = {
[chainId: string]: Export[];
};
As you see the second format include the previous. While in most case you'll need the single export where your application will support only one network, there are case where your app would want to support multiple networks at once. This second format allow for that.
Furthermore as hardhat support multiple network configuration for the same network (rinkeby, mainnet...), the export-all format will contains each of them grouped by their chainId.
Note: from v0.10.4 the old multi-export down is no more:
`
export type MultiExport = {
[chainId: string]: {[name: string]: Export};
};
For both --export and --export-all, if the extension ends in .ts it will generate a typescript file containing the contracts info.
---
---
As mentioned above, the deploy function can also deploy a contract through a proxy. It can be done without modification of the contract _as long as its number of constructor arguments matches the proxy initialization/update function_. If the arguments do not match, see this section below.
The default Proxy is both ERC-1967 and ERC-173 Compliant, but other proxy can be specified, like openzeppelin transparent proxies.
Code for the default Proxy can be found here.
To perform such proxy deployment, you just need to invoke the deploy function with the following options : {..., proxy: true}
See example :
`js`
module.exports = async ({getNamedAccounts, deployments, getChainId}) => {
const {deploy} = deployments;
const {deployer} = await getNamedAccounts();
await deploy('Greeter', {
from: deployer,
proxy: true,
});
};
You can also set it to proxy: " in which case the function will be executed upon upgrade.args
the field will be then used for that function instead of the contructor. It is also possible to then have a constructor with the same arguments and have the proxy be disabled. It can be useful if you want to have your contract as upgradeable in a test network but be non-upgradeable on the mainnet.
See example :
`js`
module.exports = async ({
getNamedAccounts,
deployments,
getChainId,
network,
}) => {
const {deploy} = deployments;
const {deployer} = await getNamedAccounts();
await deploy('Greeter', {
from: deployer,
proxy: network.live ? false : 'postUpgrade',
args: ['arg1', 2, 3],
});
};
The proxy option can also be an object which can set the specific owner that the proxy is going to be managed by.
See example:
`js`
module.exports = async ({getNamedAccounts, deployments, getChainId}) => {
const {deploy} = deployments;
const {deployer, greeterOwner} = await getNamedAccounts();
await deploy('Greeter', {
from: deployer,
proxy: {
owner: greeterOwner,
methodName: 'postUpgrade',
},
args: ['arg1', 2, 3],
});
};
Note that for the second invocation, this deployment will not be executed from the specified from: deployer as otherwise these tx will always fail. It will instead be automatically executed from the proxy's current owner (in that case : greeterOwner)
Now, it is likely you do not want to locally handle the private key / mnemonic of the account that manage the proxy or it could even be that the greeterOwner in question is a multi sig. As such that second invocation will throw an error as it cannot find a local signer for it.
The error will output the necessary information to upgrade the contract but hardhat-deploy comes also with a utility function for such case: deployments.catchUnknownSigner which will catch the error and output to the console the necessary information while continuing to next step.
Here is the full example :
`js`
module.exports = async ({getNamedAccounts, deployments, getChainId}) => {
const {deploy, catchUnknownSigner} = deployments;
const {deployer, greeterOwner} = await getNamedAccounts();
await catchUnknownSigner(
deploy('Greeter', {
from: deployer,
proxy: {
owner: greeterOwner,
methodName: 'postUpgrade',
},
args: ['arg1', 2, 3],
})
);
// you could pause the deployment here and wait for input to continue
};
#### When the constructor and init functions are different
When the constructor and proxy have different signatures, you will not be able to use the top level args property. Instead you can use the execute property of proxy to specify the init method and arguments. This will not try to pass any arguments to the constructor.
`typescript`
const deployed = await deploy('YourContract', {
from: deployer,
proxy: {
execute: {
init: {
methodName: 'initialize',
args: ['arg1', 'arg2'],
},
},
proxyContract: 'OpenZeppelinTransparentProxy',
},
log: true,
autoMine: true,
});
#### Proxy deployment options
The full proxy options is as follow:
`ts
type ProxyOptionsBase = {
owner?: Address; // this set the owner of the proxy. further upgrade will need to be executed from that owner
upgradeIndex?: number; // allow you to breakdown your upgrades into separate deploy script, each with their own index. A deploy call with a specific upgradeIndex will be executed only once, only if the current upgradeIndex is one less.
proxyContract?: // default to "EIP173Proxy". See below for more details
string | ArtifactData;
viaAdminContract?: // allow to specify a contract that act as a middle man to perform upgrades. Useful and Recommended for Transparent Proxies
| string
| {
name: string;
artifact?: string | ArtifactData;
};
};
export type ProxyOptions =
| (ProxyOptionsBase & {
methodName?: string; // method to be executed when the proxy is deployed for the first time or when the implementation is modified. Use the deployOptions args field for arguments
})
| (ProxyOptionsBase & {
execute?:
| {
methodName: string; // method to be executed when the proxy is deployed for the first time or when the implementation is modified.
args: any[];
}
| {
init: {
methodName: string; // method to be executed when the proxy is deployed
args: any[];
};
onUpgrade?: {
methodName: string; // method to be executed when the proxy is upgraded (not first deployment)
args: any[];
};
};
});
`
The proxyContract field allow you to specify your own Proxy contract. If it is a string, it will first attempt to get an artifact with that name. If not found it will fallback on the following if
it matches:
- EIP173Proxy: use the default Proxy that is EIP-173 compliant
- EIP173ProxyWithReceive: Same as above except that the proxy contains a receive hook to accept empty ETH payment.
- OpenZeppelinTransparentProxy: Use Openzeppelin Transparent Proxy (copied from openzeppelin repo, see code here)DefaultProxyAdmin
When this option is chosen, the is also used as admin since Transparent Proxy kind of need an intermediary contract for administration. This can be configured via the viaAdminContract option. Note that the DefaultProxyAdmin is slightly different than the one used by openzeppelin as it allow you to set a different owner than msg.sender on first deploy, something openzeppelin version do not allow, see : https://github.com/OpenZeppelin/openzeppelin-contracts/issues/2639
- OptimizedTransparentProxy: This contract is similar to above, except that it is optimized to not require storage read for the admin on every call.
The deployments field also expose the diamond field: hre.deployments.diamond that let you deploy Diamonds in an easy way.
Instead of specifying the facets to cut out or cut in, which the diamond contract expects, you specify the facets you want to end up having on the deployed contract.
This declarative approach allow you to focus on what you want instead of how to do it.
diamond.deploy expect the facet as names. The names represent contract to be deployed as facet. In future version you ll be able to specify deployed contract or artifact object as facet.
To deploy a contract with 3 facet you can do as follow :
`js`
module.exports = async ({getNamedAccounts, deployments, getChainId}) => {
const {diamond} = deployments;
const {deployer, diamondAdmin} = await getNamedAccounts();
await diamond.deploy('ADiamondContract', {
from: deployer,
owner: diamondAdmin,
facets: ['Facet1', 'Facet2', 'Facet3'],
});
};
if you then later execute the following script:
`js`
module.exports = async ({getNamedAccounts, deployments, getChainId}) => {
const {diamond} = deployments;
const {deployer, diamondAdmin} = await getNamedAccounts();
await diamond.deploy('ADiamondContract', {
from: diamondAdmin, // this need to be the diamondAdmin for upgrade
owner: diamondAdmin,
facets: ['NewFacet', 'Facet2', 'Facet3'],
});
};
Then the NewFacet will be deployed automatically if needed and then the diamondCut will cut Facet1 out and add NewFacet.
Note that if the code for Facet2 and Facet3 changes, they will also be redeployed automatically and the diamondCuts will replace the existing facets with these new ones.
Note that the diamond has 3 facet added by default. These facets are used for ownership, diamondCut and diamond loupe.
The implementation is a sligthly modified version of the reference implementation by Nick Mudge.
The only difference is the custom constructor that allow multiple initialization, used to allow the default ERC165 facet to be initialised along your custom initialization function.
Like normal proxies you can also execute a function at the time of an upgrade.
This is done by specifying the execute field in the diamond deploy options :
`js`
diamond.deploy('ADiamondContract', {
from: deployer,
owner: diamondAdmin,
facets: ['NewFacet', 'Facet2', 'Facet3'],
execute: {
methodName: 'postUpgrade',
args: ['one', 2, '0x3'],
},
});
There are more options, to be described later...
You can continue using the usual test task:
hardhat test
Tests can use the hre.deployments.fixture function to run the deployment and snapshot it so that tests don't need to perform all the deployment transactions every time. They can simply reuse the snapshot for every test (this leverages evm_snapshot and evm_revert provided by both hardhat and ganache). You can for example set them in a beforeEach.
Here is an example of a test :
`js
const {deployments} = require('hardhat');
describe('Token', () => {
it('testing 1 2 3', async function () {
await deployments.fixture(['Token']);
const Token = await deployments.get('Token'); // Token is available because the fixture was executed
console.log(Token.address);
const ERC721BidSale = await deployments.get('ERC721BidSale');
console.log({ERC721BidSale});
});
});
`
Tests can also leverage named accounts for clearer test. Combined with hardhat-deploy-ethers plugin, you can write succint test :
`js
const {ethers, getNamedAccounts} = require('hardhat');
describe('Token', () => {
it('testing 1 2 3', async function () {
await deployments.fixture(['Token']);
const {tokenOwner} = await getNamedAccounts();
const TokenContract = await ethers.getContract('Token', tokenOwner);
await TokenContract.mint(2).then((tx) => tx.wait());
});
});
`
---
Furthermore, tests can easily create efficient fixture using deployments.createFixture
See example :
`jscreateFixture
const setupTest = deployments.createFixture(
async ({deployments, getNamedAccounts, ethers}, options) => {
await deployments.fixture(); // ensure you start from a fresh deployments
const {tokenOwner} = await getNamedAccounts();
const TokenContract = await ethers.getContract('Token', tokenOwner);
await TokenContract.mint(10).then((tx) => tx.wait()); //this mint is executed once and then will ensure it is snapshotted`
return {
tokenOwner: {
address: tokenOwner,
TokenContract,
},
};
}
);
describe('Token', () => {
it('testing 1 2 3', async function () {
const {tokenOwner} = await setupTest();
await tokenOwner.TokenContract.mint(2);
});
});
While this example is trivial, some fixture can require several transaction and the ability to snapshot them automatically speed up the tests greatly.
---
---
---
as mentioned above, the node task is slightly modified and augmented with various flags and options
hardhat node
In particular it adds an argument --export that allows you to specify a destination file where the info about the contracts deployed is written.
Your webapp can then access all contracts information.
---
hardhat test
the test task is augmented with one flag argument --deploy-fixture that allows to run all deployments in a fixture snapshot before executing the tests. This can speed up the tests that use specific tags as the global fixture take precedence (unless specified).
In other words tests can use deployments.fixture( where specific tag only deploys the minimal contracts for tests, while still benefiting from global deployment snapshot if used.
If a test needs the deployments to only include the specific deployment specified by the tag, it can use the following :
`js`
deployments.fixture('
Due to how snapshot/revert works in hardhat, this means that these tests will not benefit from the global fixture snapshot and will have to deploy their contracts as part of the fixture call. This is automatic but means that these tests will run slower.
---
hardhat --network