Ethereum
Nethermind cuts repeated reads for state proofs and block receipts
Two November 8 commits change how storage proofs are collected and how receipts are served to peers. The changes land on the Ethereum client’s main branch.

Storage proofs take a different path
On November 8, Nethermind integrated two changes into its Ethereum execution client’s main branch. They target different read operations. The first is proof generation through the eth_getProof RPC method, especially when a request includes multiple storage slots for a contract. The second concerns responses to peers requesting transaction receipts for blocks. The changes affect how the client finds and reads the data it needs for a response. The two commits are separate and target different client components.
For eth_getProof, the previous path compared the requested keys with the path of each node visited in the storage tree. Starting at five keys, the new code sorts them on the first visit to the storage tree and preserves each key’s position in the request. Keys sharing a prefix are then contiguous in the sorted list. A binary search finds the start of that range. The collector then walks the keys that match the current path to assemble their proofs. Smaller requests retain the linear scan.
Receipts avoid an initial full decode
In the synchronization service, Nethermind first opened a full block to check whether it had a body and transactions. The receipt finder then opened the block for its own processing. The new path reads the header first and checks that the block body is available. When the transaction root indicates an empty block, the service returns an empty list. In other cases it leaves the full read to the component that finds the receipts. The change is intended to avoid decoding the same block twice on this path.
An unavailable block and an empty block still get different treatment. If the header is missing or the body is absent, the service does not return an empty list that would suggest the block contained no transactions. If the header indicates an empty block and the body is present, the service can return an empty response without retrieving receipts. This detail applies to the peer synchronization service. It does not establish the same behavior for every receipt retrieval method in the client.
Targeted tests, without production measurements
The repository adds test cases comparing batched storage proofs with proofs requested individually. The internal sort preserves each key’s association with its original request position. The test code also covers repeated inputs. For receipts, the repository adds a benchmark scenario for blocks old enough to be outside the recent-block cache. These additions describe reproducible scenarios. The commits do not provide measurements observed in production.
November 8 marks the changes’ integration into the main branch, with neither commit announcing a distribution date. Teams building this branch can isolate the two effects by comparing proof requests with multiple keys and receipt requests for blocks outside the cache. The documents reviewed describe neither a new proof format nor a change to Ethereum’s consensus rules.
