This repo copy the USDC swap code from chainlink masterclass-3: https://cll-devrel.gitbook.io/ccip-masterclass-4/ccip-masterclass/exercise-2-deposit-transferred-usdc-to-compound-v3
The mission of this homework is to have a well-rounded gas test for the ccipReceive() function, in which "the CrossChainReceiver.sol smart contract will swap received USDC tokens for Compound's mock USDC tokens, deposit those mock tokens into the Compound V3 by calling the Comet.sol smart contract's supply function and CrossChainReceiver.sol smart contract should get COMP tokens in return" )
The call flow: USDC-Sender-A -> Router -> Receiver (1. first call swap() in exchange of cUSDC; 2. call supply.sol).
We test this two-phase receiver logic within Foundry.
There are the contracts involved during our gas test:
- Internal Contracts:
CrosschainReceiver.solSwapTokenUSDC.sol
- External Contracts interact with the
ccipReceive()funciton:ERC20(USDC)ERC20(cUSDC)comet.sol
Generally, we have following methods to evaulate:
- Deploy our internal contracts within testnet and interact with external contracts.
- Deploy our internal contracts and real external contract within the local network.
- Deploy our internal contracts locally and mock our external contracts within the local network.
In method one, as we donot fully own the external contracts, we can not range all interaction results, and the test times is highly constrained by the amount of testnet-tokens. Besides, the test time within real testnet like sepolia will be quite long.
In method two, the external contracts(especially the comet.sol contract, which has a list of sub-contract to interact, composing the full compound V3 protocol) are defficult to evaluate.
In method three, we can mock the execution result of external contracts with simplier simulated contracts, and we can mock the gas cost by using data from other unite tests and on-chain data. For example, this site logs the Comet.supply()'s gas cost within real world.
To simulate the external contracts, we mock those contract with real-world gas consumption:
| src/mocks/mockComet.sol:MockComet contract | | | | | |
|--------------------------------------------|-----------------|-------|--------|-------|---------|
| Deployment Cost | Deployment Size | | | | |
| 135619 | 414 | | | | |
| Function Name | min | avg | median | max | # calls |
| supply | 88386 | 88386 | 88386 | 88386 | 1 |
| src/mocks/mockERC20.sol:ERC20Mock contract | | | | | |
|--------------------------------------------|-----------------|-------|--------|-------|---------|
| Deployment Cost | Deployment Size | | | | |
| 673857 | 3672 | | | | |
| Function Name | min | avg | median | max | # calls |
| approve | 46272 | 46272 | 46272 | 46272 | 1 |
| balanceOf | 562 | 562 | 562 | 562 | 2 |
| mint | 34049 | 48261 | 51149 | 51149 | 6 |
Then we test the ccipReceive Function under three different condition:
| src/CrosschainReceiver.sol:CrossChainReceiver contract | | | | | |
|--------------------------------------------------------|-----------------|--------|--------|--------|---------|
| Deployment Cost | Deployment Size | | | | |
| 1781129 | 8590 | | | | |
| Function Name | min | avg | median | max | # calls |
| allowlistSender | 46203 | 46203 | 46203 | 46203 | 3 |
| allowlistSourceChain | 46169 | 46169 | 46169 | 46169 | 3 |
| ccipReceive | 214251 | 315914 | 264805 | 468688 | 3 |
| setSimRevert | 45840 | 45840 | 45840 | 45840 | 1 |
- If the receive function works smoothly with no error, the estimated gas would around
214251. - If it revoke instantly after the it got the message, the estimated gas would around
264805. - If it revoke until the last step (the
Comet.Supply()function), the estimated gas would around468688.
So to secure the gaslimit should set to 1.1 x 468688.