chore(payload): fix withdrawals field pre-shanghai in Ethereum payload

[fgimenez]

We were getting a hive error Unexpected Failures: ['GetPayloadBodiesByRange (Empty Transactions/Withdrawals) (Paris) (reth)'] in the ethereum-withdrawals suite after #12780, like in https://github.com/paradigmxyz/reth/actions/runs/11982076699/job/33409959229, withdrawals field should be None pre-shanghai

[fgimenez]

successful hive run from this branch https://github.com/paradigmxyz/reth/actions/runs/11995148078

[mattsse]

ah sigh, I missed this in #12780

[hai-rise]

Totally missed commit_withdrawals returning None pre-Shanghai. Sorry guys 🫠.

Also surprised Hive would expect an overwritten None when clients like Geth reject such a payload with an error. Will have a look at Hive if I can 🙏.

[fgimenez]

Totally missed commit_withdrawals returning None pre-Shanghai. Sorry guys 🫠.

sure np at all, thanks for your contributions!

Also surprised Hive would expect an overwritten None when clients like Geth reject such a payload with an error. Will have a look at Hive if I can 🙏.

in fact the test is exercising engine_getPayloadBodiesByRangeV1, the error we were getting was 2024-11-24T00:36:49.8832431Z FAIL (GetPayloadBodiesByRange (Empty Transactions/Withdrawals)): Unexpected payload body on EngineGetPayloadBodiesV1 at index 0: incorrect withdrawals: want: null, got: [] the test is defined here if you want to take a look https://github.com/ethereum/hive/blob/master/simulators/ethereum/engine/suites/withdrawals/tests.go#L678-L709

[hai-rise]

@fgimenez Thank you for the info and fixing the regression!

I think I understand my confusion now. In Geth Withdrawals can be nil in both new payload requests and built block bodies (both use ExecutableData):
https://github.com/ethereum-optimism/op-geth/blob/23fc61420c904ae9d7b8e5978b39bedd469cd283/beacon/engine/types.go#L77-L92

That's why they confidently forward it from the payload attributes to the built block body:
https://github.com/ethereum-optimism/op-geth/blob/23fc61420c904ae9d7b8e5978b39bedd469cd283/beacon/engine/types.go#L313

However, in Alloy, we parse null withdrawals as an empty Vec in new payload requests:
https://github.com/alloy-rs/alloy/blob/1ef28dfb6b97c580a08e5750569e385555c0747c/crates/rpc-types-engine/src/payload.rs#L214

While Reth does Option<Withdrawals> for BlockBody.

reth/crates/primitives/src/block.rs

Line 602 in 02824da

pub withdrawals: Option<Withdrawals>,

Ideally, ExecutionPayloadV2 would also be Option for us to forward and match the value between the payload request and the built block as Geth does. So we'd forward a None withdrawals correctly in the above test, instead of parsing null input as [] then need to check Shanghai to revert to null.

A change in Alloy is very noisy so may not worth it to be honest 😅. What do you think?