@fradamt This is a good point. Native rollups w/o ZKPs are essentially a parallel gas limit boost. Can be done on a separate CPU core.
Yet each new tx comes with a merkle proof which is a bandwidth burden. Better to just increase L1 gas up to where single core block execution is limiting
For native rollups, BLOCKHASH opcode use is not secured by the state root. It is implicit state, which L1 doesn't know
One solution is to have an L1 contract with a 256 long ring buffer for each rollup. After EXECUTE returns valid, the L2 blockhash buffer is updated
@0xalizk Very nice
> @drakefjustin: a stateless format is bottleneck to fusaka native rollups
Here's a format that could work/accelerate:
https://t.co/if59zqHguz
FOCILIS - an extension to FOCIL that uses a semaphore-like construction.
This allows IL publishers to be truly honest about what they see in the mempool.
I think the ZKPs are probably mature enough. What do you think?
https://t.co/gJH9q9AmOQ
@FUCORY@lightclients@nero_eth In this case (prototyping a browser extension), the standard means you don't have to do the work thinking through what data is required.
You can just get to hacking
Because its a useful abstracrion (get all state that a block touches, and the merkle proofs for that state)
@FUCORY@lightclients@nero_eth Etherscan themselves could send you the block prestate (so you can check the data they present). Or another provider could
A standard means they can run any client they want in the backend
A standard solves the problem "how to get a provider on board?": nodes do it by default
@FUCORY@lightclients What I mean is that verkle can enable new things (use cases / paradigms)
But it may be that after verkle, the things to not appear. See blake2f.
If verkle truly leads to useful things, those things can actually be made today (to prove pmf). They will just be a bit heavy.
@FUCORY@lightclients IMO verkle ideas can be tried now
Broke: verkle fixes this
Bespoke: merkle fixes this
Merkle version is just a bit more bandwidth. But that's OK to see if an idea has pmf
@FUCORY@lightclients@nero_eth There's also a rust app called archors that makes and verifies payload. So you can test the idea by hammering a local node of any client.
Browser extension identifies the block number on etherscan -> get block -> getRequiredBlockState, verify -> extension checks etherscan data
@FUCORY@lightclients@nero_eth Some standards work I did in the execution apis repo might be relevant here. PR455 eth_getRequiredBlockState
It's all the prestate relevant for a block, accompanied by merkle proofs. ~167 KB/Mgas
@0xsileo Suppose people decided that they want archive nodes to be a bit bigger to gain the ability of address introspection.
I perceived a need and so wrote EIPs for this to help out. But I do not say or even know whether people should or shouldn't want this.
@0xsileo I have struggled with this idea though. There are ideals and realities. It needs to be possible for end users to have full control. But how easy should it be, if there are many orders of magnitude more users who have a less fastidious approach? Is the end game snarks on phones?
@gakonst Associated eth_getRequiredBlockState method.
Outlines a format for the minimum required data at the block level. So data is small and reproducible.
https://t.co/3frrbEcb4T