January 14th tl;dc (too long, didn’t call)
Disclaimer: This is a summary of the topics discussed in the Eth1.x research call. It does not represent finalized plans or commitments for network upgrades. The main topics covered were the advantages of switching to a binary trie structure, transition strategies and challenges, “merklizing” contract code, and chain pruning and historical data distribution.
– There will be a small 1.x research summit on March 7-8 following EthCC.
– The next call is tentatively scheduled for the first or second week in February.
– This EIP improves the network protocol for transaction propagation and aligns with the research goals.
Binary Trie size savings:
– Transitioning to a binary trie structure could reduce witness sizes by approximately 3.75x in theory.
– In practice, the reduction might only be about half, depending on various factors.
– Switching to a binary trie format would bring witness sizes to ~300-1400kB, down from ~800-3,400kB in the hexary trie.
Making the switch:
– Two strategies were discussed: progressive transition and compute and clean-cut.
– Progressive transition involves gradually migrating the state trie to a binary format, while accounts would need to be “poked” for the new trie format.
– Compute and clean-cut involves a one-time transition during a hard fork where participants in the network switch to the new trie format together.
– Both strategies have pros and cons and require further consideration and discussion.
– Prototyping work has been done on code ‘merklization’ to split contract code into chunks for witnesses.
– This could result in a 50% reduction in witness sizes.
– The idea of creating a single global ‘code trie’ is still in early research stages.
– Changes to gas scheduling with block witnesses need to be considered, especially with regards to code chunking.
– One idea is to have the poster of a transaction pay the full cost of their transaction’s witness, with miners keeping the overpayment.
– Changes to the security model are necessary and require thorough consideration.
Pruning and data delivery:
– Stateless clients need a way to signal what data they have and lack in a network.
– This has implications for network topology and data distribution.
Note: This summary provides a brief overview of the topics discussed and does not include all details or conclusive decisions.