Constructor argument gas estimation fallback in Hardhat v2.12.0

In Hardhat v2.12.0, when constructor arguments exceed the ABI encoding threshold for inline values, the framework often fails to estimate gas correctly. This results in a gas estimate of 0, causing deployments to revert with 'out of gas' errors. Because the EVM never executes the transaction, no specific revert reason is returned from the contract.

The Root Cause

This occurs because Hardhat relies on the eth_estimateGas RPC-call before sending the transaction. If the constructor arguments are exceptionally large (e.g., massive arrays or complex structs), the underlying JSON-RPC provider may return zero or an error because it cannot simulate the execution cost of the oversized payload. Since the transaction is never actually "sent" with a valid gas limit, the Solidity logic—and its requirerequire statements—are never triggered.

Immediate Workarounds

To bypass the faulty estimation logic and surface the actual error, follow these steps in your deployment script:

1. Manually set the gas limit: Instead of letting Hardhat estimate the gas, provide a high gasLimit manually in your deploy function.

const contract = await ethers.deployContract("MyContract", args, {
  gasLimit: 5000000, // Set a sufficiently high limit manually
});

2. Use the --force flag: If using the Hardhat-ignition, some versions allow forcing a deployment to skip the estimation phase, though this is less common in standard ethers-based scripts.

Hardhat Network vs. Forked Environments

The behavior varies significantly between the local Hardhat Network and forked environments. The local network is more permissive with simulation, whereas external chain forks (like Infura or Alchemy) have strict timeouts and resource limits on gas-call simulations. When migrating between these, you should:

• Define a gasLimit in your hardhat.config.js specifically for the forked network to ensure consistency.
• Check if your provider supports debug_traceCall, which can sometimes provide more detail than standard estimation fails.

Future Framework Improvements

For Hardhat to reliably surface revert reasons, it would need to implement a "dry-run" fallback: if eth_estimateGas fails, the framework could attempt to send the transaction with a high gas limit and then parse the data field of the reverted-transaction response. However, this would consume actual gas on live networks, which is undesirable for development workflows.

Diagnostic Question: Are you encountering this specifically when deploying to a mainnet fork, or does it also occur on the local Hardhat Network node?