Sharing block templates

The recent publication of a draft Bitcoin Improvement Proposal (BIP), referred to as BIN25-2, outlines the basic protocol without addressing the optimal usage strategies or incorporating delta-encoding, which remains a topic of discussion within the community. Initial experiments on the mainnet, involving two peers with similar mempool policies, revealed minimal occurrences of out-of-mempool transactions in shared templates. These primarily included transactions that were either replaced before the generation of the template or confirmed in the previous block but not yet updated in the new template. The processing delay for just-confirmed transactions, although minimal per transaction, can accumulate significantly when multiple transactions are confirmed simultaneously.

To address these issues, leveraging the map of the last block's transactions to expedite processing and considering caching and skipping over just-replaced-by-fee (RBF) transactions have been proposed. This approach not only streamlines processing but also has potential benefits in mitigating Riard's replacement cycling attacks by reconsidering recently RBFed top-of-mempool transactions. Despite the challenges, few such cases present manageable problems.

Furthermore, enhancing relay connectivity to counteract relay censorship has been explored. A strategy involving requesting templates from feeler peers before disconnection, followed by integrating any unique transactions into one's mempool, has been identified as an effective method. This technique aims to improve relay connectivity without imposing significant ongoing burdens, thereby offering a balanced approach to maintaining network integrity in the face of adversarial actions.