Skip to content

Latest commit

 

History

History
73 lines (63 loc) · 2.96 KB

fork-test.md

File metadata and controls

73 lines (63 loc) · 2.96 KB
Raw

Fork Tests

What and Why?

Fork tests are similar to unit tests, except that it is run on a mainnet fork against already deployed contracts. It is also run as part of the CI. It helps to reduce manual testing when any changes to contracts are made. Also useful in testing contracts that depend on external contracts (like Oracle or Uniswap or Metapools).

Running Fork Tests

Option 1: Fork each time (just like CI)

  • Set PROVIDER_URL in .env file
  • Run yarn test:fork (or yarn test:fork -- filename.fork-test.js)

This forks the blockchain (using PROVIDER_URL) every time you run the command, might take a lot of time.

gitGraph
  commit id: "block n-1"
  commit id: "FORK_BLOCK_NUMBER"
  branch testRun1
  commit id: "1: local deployments"
  commit id: "1: test fixtures"
  commit id: "1: fork tests"
  commit id: "1: tests end" type: REVERSE
  checkout main
  branch testRun2
  commit id: "2: local deployments"
  commit id: "2: test fixtures"
  commit id: "2: fork tests"
  commit id: "2: tests end" type: REVERSE
  checkout main
  commit id: "block n+1"
  commit id: "block n+2"
Loading

Option 2: Nested forking (only for local/dev usage)

In cases where you don't want to fork or run the deployments again and again

  • Start the node in a separate terminal using FORK=true yarn run node
  • Run yarn test:fork in a new terminal

How it works

Basically, you start a forked node that runs the deployments and all. Next, the hardhat forks from this forked node when running the tests. Since any new deployments have already been run on the base fork, running yarn test:fork will be much faster. This can be useful when you are just writing tests and there isn't any change to contracts or deployment files.

gitGraph
  commit id: "block n-1"
  commit id: "FORK_BLOCK_NUMBER"
  branch fork
  commit id: "local deployments"
  branch testRun1
  commit id: "1: test fixtures"
  commit id: "1: fork tests"
  commit id: "1: tests end" type: REVERSE
  checkout fork
  branch testRun2
  commit id: "2: test fixtures"
  commit id: "2: fork tests"
  commit id: "2: tests end" type: REVERSE
  checkout main
  commit id: "block n+1"
  commit id: "block n+2"
  checkout fork
  commit id: "manually terminated" type: REVERSE
  checkout main
  commit id: "block n+3"
Loading

Things to know

  • Hardhat always runs the test under the default hardhat network
  • When forking.url is set for a network in hardhat.config.js file, hardhat internally spins up its own node (aka Hardhat Runtime Environment or HRE) with the network name hardhat, forks the network and snapshots it to run the tests against.
  • No changes are persisted to the actual node. All deployments and contract calls are only made on the forked node, not the actual node

Environment Variables used on CI

  • MAX_CHUNKS - The number of runners to use to run fork tests parallely on CI, hardcoded to 4 in defi.yml file
  • CHUNK_ID - Current runner ID (will be set by Github Actions), see defi.yml for the definition