The most noteworthy Ethereum Improvement Proposals in 2021

Original title: "2021: The most noteworthy Ethereum improvement proposal"

Original source: ETH Chinese website

2021 will be the most innovative year for the Ethereum protocol since 2016. Here are the EIPs worth noting this year:

The next hard fork upgrade, Berlin, will implement four EIPs:

EIP-2929: Increased gas fees for state access opcodes

EIP-2930: Optional Access List Txs

EIP-2718: Typed Transaction Envelope

EIP-2315: Simple Subroutines

EIP-2929: Increased gas fees for state access opcodes. Gas fees for storage access opcodes have long been too low, and malicious transactions that take 20-80 seconds to execute are now easily created. This EIP increases the cost of state access by about 3x.

EIP-2930: Optional access lists. About 0.18% of mainnet transactions are blocked by EIP-2929. To fill this gap, this EIP introduces a new type of transaction that can specify the storage slot to be read and then pay a 10% discounted upfront fee.

EIP-2718: Transaction envelope containing types. This EIP specifies a new format for transactions that starts with a "discriminator" to allow clients to decipher different transactions. EIP-2930 will use this EIP first, and then possibly EIP-1559. New format = type || payload.

EIP-2315: Simple Subroutines. The last EIP included in the Berlin hard fork upgrade: adding support for subroutines. Traditionally, contracts that want to emulate this functionality need to handle both regular stack data and return addresses. This EIP adds a new stack that is only used to track return addresses.

After the Berlin fork, there will be many unstable factors. If the Ice Age does not extend in the next fork, the fork will start in early July. According to the recent AllCoreDev meeting, there may be another fork this summer. There are many candidate EIPs for this hard fork, the most famous of which is EIP-1559.

EIP-1559: Fee market changes for Eth1. Gas fees are divided into two parts: base fee and tip. The base fee is destroyed, while the tip is sent to miners. The base fee is dynamically determined by the protocol based on demand. This can avoid some fee peaks and improve the user experience when paying gas fees.

(Translator’s note: The author has inserted the latest status update of EIP-1559 by Tim Beiko, the main promoter of EIP-1559)

Details: https://twitter.com/TimBeiko/status/1349027063903121408

In my opinion, I don't think 1559 is ready to go live right now. Here is a summary of some of the thoughts I shared on dicord:

Not all clients have an implementation of 1559, and no client has a final version;

We haven’t solved the issues raised at the AllCoreDevs meeting (i.e. testing large blocks with large states). It will take weeks at best to achieve this.

1559 will impact many other projects: any project that involves signing transactions, relies on JSON RPC responses to blocks/transactions, etc. We still need to talk extensively about these impacts.

The Berlin hard fork is about to start. It is not a good choice to put down all the work at hand to promote 1559.

EIP-2982: Phase 0 of Eth2. For the first time since EIP-1011, eth1 and eth2 are combined in the EIP repository. This EIP is just the beginning of reintroducing eth2 governance/standardization into eth1. The main changes it brings are: New ETH issuance in eth2.

EIP-?: Merge. On a more optimistic note, eth1 -> eth2 is expected to merge in 2021 and remove PoW consensus. However, in the coming months, research and development work will increase significantly. It is estimated that this EIP will be carried out in the next 18 months.

EIP-3102: Binary Trie Structure. This EIP will move from the current hexa-state tree to a binary trie structure. This will make Merkle proofs more efficient. The migration will likely take up a lot of resources, so it’s best to do it before the merge.

EIP-2937: SETINDESTRUCTIBLE opcode. This opcode will cause any subsequent SELFDESTRUCTs to act as noops in the current frame. This will help to better demonstrate the reliability of the contract. This is also the first thing to do to support account abstraction.

EIP-2938: Account Abstraction. Allows contracts to become high-level accounts that pay transaction fees and initiate transaction execution. This provides a flexible framework for determining the conditions for the validity of any transaction. Originally proposed in EIP-86, this EIP provides a more specific path for implementation on the mainnet.

EIP-?: EVM384. While this is not yet an official EIP, it is moving in that direction. The proposal proposes adding native 384-bit arithmetic operators (and likely other widths) to support efficient implementation of new cryptographic systems in the EVM.

EVM384 (Contents). With the rapid innovation of modern cryptographic systems, EVM384 provides a framework for Ethereum to support these systems without the need for hard forks and precompiles for each cryptographic system. Instead, anyone can implement it as an EVM contract.

EIP-2537: BLS12-381 precompile. Whether this proposal is still needed after EVM384 is available is still under discussion. Although EVM384 is a better long-term solution, BLS is an advanced proposal in the Ethereum ecosystem (given eth2) and is worth precompiling locally.

Batched transactions. There are many EIPs that aim to replace meta-transactions with L1 primitives. Expect significant progress in this area this year. Such EIPs are:

EIP-2711: Sponsored, expiring and batch txs

EIP-2733: Transaction Packaging

EIP-2803: Expensive Transactions

EIP-3074: Sponsored Tx Precompile

EIP-2464: eth/65: Transaction Broadcast and Retrieval. This proposal significantly reduces the bandwidth used by eth1 clients by broadcasting transaction hashes to a majority of peer nodes instead of broadcasting full transactions. Geth plans to stop supporting < eth/65 sometime this year.

2021 will bring many changes to the Ethereum protocol, far more than what I have listed. The Eth2 and Stateless Ethereum teams will have their own plans for this year.