feat(`cheats`): add RLP related cheats

[0xClandestine]

Component

Forge

Describe the feature you would like

It would be useful imo to add RLP encoding/decoding related cheats, mirroring the functionality already available in cast via cast to-rlp and cast from-rlp. This would allow contracts to easily manage RLP data in tests without relying on external libraries or tooling.

/// @dev RLP encodes a list of `data` into an rlp payload.
function toRlp(bytes[] calldata data) public pure returns (bytes memory);
/// @dev RLP decodes an `rlp` payload into a list of `data`.
function fromRlp(bytes calldata rlp) public pure returns (bytes[] memory data);

Rather than bytes[] calldata data it may be cleaner to simply take in bytes calldata abiEncoded so usage becomes:

bytes memory rlpPayload = vm.toRlp(abi.encode(item0, item1, item2...));
(bytes item0, uint256 item1, address item2, ...) = abi.decode(vm.fromRlp(rlpPayload), (bytes, uint256, address);

Worth noting the abi.encode pattern might be annoying if you don't want to parse all elements.

Additional context

No response

[grandizzy]

I think this makes sense 👍 cc @zerosnacks wdyt?

[zerosnacks]

No objections, makes sense

[0xClandestine]

Some additional ideas:

It may be useful to expose functions that parse specific elements by index from a RLP-encoded list. This is particularly relevant for decoding structures like Ethereum block headers, which typically contain uint, address, bytes32, and a string (extraData). Also complements the recent new cheat vm.getRawBlockHeader.

function rlpUint(bytes calldata rlp, uint256 index) external pure returns (uint256);
function rlpAddress(bytes calldata rlp, uint256 index) external pure returns (address);
function rlpBytes32(bytes calldata rlp, uint256 index) external pure returns (bytes32);
function rlpBytes(bytes calldata rlp, uint256 index) external pure returns (bytes memory);
function rlpString(bytes calldata rlp, uint256 index) external pure returns (string memory);

Some utility methods for list construction could be nice as well, worth noting this functionality could be added via forge-std, but native implementations may be more efficient (if that matters here).

function rlpAppend(bytes memory rlp, uint256 value) external pure returns (bytes memory);
function rlpAppend(bytes memory rlp, address value) external pure returns (bytes memory);
function rlpAppend(bytes memory rlp, bytes32 value) external pure returns (bytes memory);
function rlpAppend(bytes memory rlp, bytes memory value) external pure returns (bytes memory);
function rlpAppend(bytes memory rlp, string memory value) external pure returns (bytes memory);

[zerosnacks]

I would like to first see an implementation of the initial proposal and evaluate traction before expanding to the possible additional utilities as every new cheatcode adds overhead, marking this as good first issue

[qiweiii]

hi, i made a pr to add these #11232

[DaniPopes]

What is the use case for this? "without relying on external libraries or tooling" is not generally a compelling argument for upstreaming stuff into cheatcodes

[0xClandestine]

What is the use case for this? "without relying on external libraries or tooling" is not generally a compelling argument for upstreaming stuff into cheatcodes

@DaniPopes That's fair, the stronger case here is that we already have vm.getRawBlockHeader() returning an RLP-encoded header, but no native way to parse it. Current onchain RLP implementations are very inefficient, which means slower test compile and runtime speeds.

Additionally, some of the most important protocols in the industry (light clients, bridges, etc.) rely on historical state proofs. They would benefit from faster testing/CI cycles and the ability to differential test their onchain implementations without FFI.

Right now the fastest way to parse block headers and trie proofs is via vm.ffi("cast block --json") or vm.ffi("cast proof <acct> --json") (pseudocode) plus manual vm.json parsing, which is cumbersome given a block header alone has 21 fields and growing (some of which aren't accessible via cheats/solidity elsewhere).

Adding RLP-related cheats makes RLP a first-class citizen (which it should be imo), providing a generic alternative to future specialized cheats for parsing block headers, account data, and other trie values.