Adds backwards compatibility in UsingPrecompile contract.

[soloseng]

Description

Updated UsingPrecompile contract to maintain backwards compatibility after L2 migration.

Fucntions that maintained backwards compatibility query the required information from the EpochManager contract instead of the precompiles.

• Remove onlyL1 in validators.sol

Other changes

Was required to add the list of elected validators to the Epoch struct of the EpochManager contract to keep the historical list of elected validators at a given epoch number.

Tested

Added unit test.

Related issues

• Fixes https://github.com/celo-org/celo-blockchain-planning/issues/642

Backwards compatibility

for some select functions

[openzeppelin-code]

Adds backwards compatibility in UsingPrecompile contract.

Generated at commit: 50bec5c8a89c6e7d9026818ef1d591cf0e891b53

🚨 Report Summary

	Severity Level	Results
Contracts	Critical High Medium Low Note Total	3 2 0 14 43 62
Dependencies	Critical High Medium Low Note Total	0 0 0 0 0 0

---

For more details view the full report in OpenZeppelin Code Inspector

[soloseng]

As of 50bec5c
I was able to add the required functions to preserve backwards compatibility to the UsingRegistry contract.

Testing is not fully implemented for registry functions, but the plan was to use EpochManager_WithMock.sol for testing needs and if possible remove MockEpochManager.sol

stCELO only uses numberValidatorsInCurrentSet ; validatorSignerAddressFromCurrentSet but I also added compatibility for the functions that perform the same query using specific blocknumber incase other contracts in the CELo ecosystem use them.

To get the epoch info based on block number, I used binary search to avoid looping through all the epoch 1x1. This was required for getEpochNumberOfBlock(); validatorSignerAddressFromSet()

We could further improve the compute cost by introducing a mapping of blocksToEpoch,

mapping(uint256 => uint256) internal blockToEpoch;

but that would be a more complex change and increase the gas cost of processing an epoch as we'd have to store this info during processing.

NOTE: Not all functions have backwards compatibility.

[m-chrzan]

Looks good, but there's a few, potentially very impactful, gas optimizations possible. There's no need to read the entire elected array into memory when we only need one field of the Epoch struct, or just one of the elected addresses. These should be fairly simple changes that would reduce the number of SLOADs by a factor of ~100 where relevant.

[martinvol]

I don't think we need to fix all precompiles in all contracts