Content
@
0 reply
0 recast
0 reaction
✳️ dcposch on daimo
@dcposch.eth
Happy New Year! We're prototyping bundle compression to make contract wallets efficient: 4337 on L2, even cheaper than EOA. Here’s how it works & how you can use it: https://daimo.super.site/blog/bulk
4 replies
11 recasts
72 reactions
Wilson Cusack
@wilsoncusack
Nice! It seems like Vitalik’s focus historically has been on amortized compression from signature aggregation. Thoughts on this?
2 replies
0 recast
4 reactions
✳️ dcposch on daimo
@dcposch.eth
Signature aggregation is useful too, but 1. Doesn't remove need for bundle compression. The UserOps themselves are way too big ABI encoded > fix by not calling EntryPoint directly. 2. Doesn't help in short term. Need a lot of volume in that specific aggregation scheme first. 2. Difficult UX tradeoff with latency.
1 reply
0 recast
2 reactions
✳️ dcposch on daimo
@dcposch.eth
Sig aggregation requires zk for use with passkeys or device keys (as opposed to BLS keys). Long term, we'll have signature aggregation via SNARK. Discussed with Barry Whitehat. For us, useful once SNARK proving is faster and we're averaging multiple transfers per second.
1 reply
0 recast
0 reaction
Wilson Cusack
@wilsoncusack
What are the numbers on snark proving gas? Thought similar to passkeys today.
2 replies
0 recast
0 reaction
Wilson Cusack
@wilsoncusack
If we’re talking L2s, the abi encoding shouldn’t be affecting as long as you’re getting accurate compression pricing. Will be better on opstack soon.
1 reply
0 recast
0 reaction
Wilson Cusack
@wilsoncusack
Rest of the points make sense, thanks!
1 reply
0 recast
0 reaction
✳️ dcposch on daimo
@dcposch.eth
I saw Proto's PR that charges txs based on compressed calldata will help! but default 4337 userop encoding is still pretty inefficient. we're getting a lot of mileage from templating many of the fields out entirely--callGasLimit, verificationGasLimit etc
0 reply
0 recast
0 reaction
