tl;dr: During the last couple of days, some L2s seen that their deployments on Sepolia did not work correctly anymore. This is because of EIP-7594 which modifications the format for proofs. In anticipation of Fusaka, we urge all blob originators to replace their software program to create Cell Proofs as a substitute of blob proofs.
Background
One underdiscussed side of EIP-7549 (PeerDAS) is that it modifications the format for proofs from blob proofs to cell proofs. This permits for downloading a particular a part of the blob as a substitute of the entire blob for knowledge availability sampling.
This alteration may break person functions that ship blob transactions. Already signed transactions are nonetheless legitimate although, they only have to recompute the cell proofs. Some shoppers (most notably go-ethereum) will do that through the RPC on eth_sendTransaction and eth_sendRawTransaction (1). This conversion from blob proofs to cell proofs takes about one second, so we encourage blob transaction originators to maneuver to cell proofs to be able to scale back overhead on the RPC degree.
Transactions which are within the txpool on the time of the fork can be dropped by some implementations, whereas different implementations will convert them to cell proofs. So it will be prudent to resend your transactions with cell proofs shortly after the fork, if they don’t seem to be being included by the chain. Some implementations enable for the distribution of blob proof transactions up to some minutes after the exhausting fork on the networking layer for stability functions.
Actionable modifications
If you’re a blob transaction originator (e.g. an L2), it is best to replace your transaction sending code to create cell proofs.
All main consumer libraries expose performance to create these proofs through ComputeCellsAndKZGProofs() which is obtainable in all main languages (2). Utilization examples can be present in all main languages within the consumer libraries (3).
Outlook
We’ll attempt to talk these modifications that impression customers extra clearly through the Ethereum weblog going ahead and attempt to do extra group outreach to stop customers of Ethereum from feeling blindsided by modifications within the protocol.
We additionally encourage L2s and different entities that closely rely on the Ethereum roadmap to comply with the ACD course of and to have interaction extra immediately with the group. We’d additionally prefer to encourage groups to deploy their contracts and take a look at infrastructure on the devnets shortly earlier than we transfer to the primary testnets.
One other useful gizmo for preserving your modifications updated is to make use of the Ethereum bundle supplied by Kurtosis which can be utilized to create native networks with the most recent specs (4).
Whereas it’s unlucky that blob originators discovered this late within the exhausting fork course of, it additionally exhibits that the testnet course of works very properly and these points are caught lengthy earlier than they might ever seem on mainnet.
(1): go-ethereum will solely do the conversion on eth_sendRawTransaction from present grasp and v1.16.5 onwards
(2): see https://github.com/search?q=repo%3Aethereum%2Fc-kzg-4844%20ComputeCellsAndKZGProofs&type=code
(3) Instance for go-ethereum: https://github.com/ethereum/go-ethereum/blob/7c107c2691fa66a1da60e2b95f5946c3a3921b00/crypto/kzg4844/kzg4844_test.go#L194
(4) Ethereum bundle for kurtosis: https://github.com/ethpandaops/ethereum-package
