make mintTransaction called via CrossDomainMessenger

[blockchaindevsh]

This PR makes the deposit tx corresponding to mintTransaction triggered via L2CrossDomainMessenger.relayMessage so that if the execution failed on L2, it can be retried later.

[blockchaindevsh]

Needs fix, will create a new pr later.

[blockchaindevsh]

reopen , I was hesitant for the msg.sender parameter which is used as sender on L2: for OP standard bridge, it's the bridge address and checked in bridge function via onlyOtherBridge; for mintTransaction, it'll be the portal address.

Other candidate values may be msg.origin/QKC_DEPOSITOR_ACCOUNT, but looks like it doesn't really matter for us.

[qzhodl]

Another approach is to introduce a new method, sendQKCTransferMessage, in L1CrossDomainMessenger.sol, which would call portal.depositTransactionChecked which can only be called by l1CrossDomainMessenger. And then we can just delete mintTransaction.

[blockchaindevsh]

Good point, now the call direction is L1CrossDomainMessenger->OptimismPortal2, not the reverse. But I kept the naming mintTransaction.

[qzhodl]

It appears that mintTransaction and depositTransaction share most of the same code. We could implement a shared _depositTransaction internal function containing the common logic, while allowing mintTransaction and depositTransaction to each handle their specific access control requirements.

[blockchaindevsh]

I'd like to keep depositTransaction untouched, otherwise it'll be a pain to merge in future.

[qzhodl]

If the method is intended solely for minting QKC, we should omit this parameter

[blockchaindevsh]

It tries to be flexible, e.g., it's possible to mint and swap in a single L2 transaction if we want it in future.

[qzhodl]

If we're using RECEIVE_DEFAULT_GAS_LIMIT, is there still a need to include the _minGasLimit parameter?

[blockchaindevsh]

Like the existing sendMessage, it's designed to be flexible, so that user can choose a proper _minGasLimit for the call.

[qzhodl]

I’m wondering if you’ve already tested whether the failed transaction can be replayed on the devnet?

[qzhodl]

I’m wondering if you’ve already tested whether the failed transaction can be replayed on the devnet?

@iteyelmp will try to test it on the devnet to see if it works.

[qizhou]

Another option is to change L1StandardBridge and move all mint logic to that so that CrossDomainMessager and Portal can be unchanged?

[blockchaindevsh]

Another option is to change L1StandardBridge and move all mint logic to that so that CrossDomainMessager and Portal can be unchanged?

portal.depositTransaction charges actual msg.value, so portal.mintTransaction is still needed.

[iteyelmp]

Summary: A devnet has been deployed based on this contract. These funds can currently be retrieved successfully by manually calling L2CrossDomainMessenger.relayMessage, but when doing so, the default estimated gas limit is incorrect — it estimates around 40,000, while the actual cost is about 70,000–80,000.

[qzhodl]

@qizhou noted that the previous CGT design also lacked tx-replay support.

To minimize contract changes and keep upstream merges straightforward, we’ll defer merging this PR into the branch for now. Instead, @iteyelmp, we can surface a notice in the migration UI:

Important Note: Do not send QKC to an L2 address that cannot receive tokens—for example, an address enabled with EIP-7702 but unable to accept native tokens. Otherwise, your funds will be permanently lost.