Scott Sunarto
@scott
!! farcaster exclusive content !! i'm building a simplified rollup from scratch to demystify ✨rollups ✨ kinda crazy how a lot of "web3 devs" see rollups as a blackbox that "scales" things. hopefully this changes it. just finished writing the execution layer. https://github.com/smsunarto/cookie-clicker-rollup
8 replies
14 recasts
61 reactions
Scott Sunarto
@scott
challenges ahead: - replicating the state processor in pure solidity. i've been very cautious of how i do serde in the EL state processor to not use non deterministic stuff and consistently use RLP/keccak256 to make sure i can easily replicate it in solidity.
1 reply
0 recast
1 reaction
Scott Sunarto
@scott
- writing a _reliable_ sequencer might be harder than it seems. need to make sure transaction are not stuck in the mempool... - i really want to do a single shot fraud prover, but worried abt running out of gas if i replay an entire block. might have to store intermediate state root.
1 reply
0 recast
0 reaction
Andrew Huang
@ahuang
curious why txs would be stuck in the mempool? just queue implementation or actual basefee / validity conditions?
1 reply
0 recast
1 reaction
Scott Sunarto
@scott
i'd imagine this would rarely happen these days, but in times of L1 congestion there are often instances where transactions are just "stuck" in the mempool and are "soft dropped". you'd have to replay the tx with the same nonce to bump it up
1 reply
0 recast
0 reaction
Andrew Huang
@ahuang
Ah L1 fee bumps… probably more of a transaction relay issue than sequencer issue (client need to opt-in to higher fee)
1 reply
0 recast
1 reaction
