fredo
commented
1 year ago Summary
Optimism changes in a way that messages have to be proved first (on L1) before the 7 day window starts. After the 7 days relayMessage can be called. This also implies that without calling prove, the 7 day window never starts. This means that it takes an L1 transaction to start the 7 day window. @GabrielBuragev can you please provide the links to the documentation.
This is very unfortunate as it complicates our timing assumptions. This leads to the fact that for proper solutions we are required to change our contracts. This should be it's stand alone issue, to research and decide on a longterm solution.
Short term solution
It would be not feasible if the agent would always call prove for every transfer even in the happy case. However, when an agent is part of the challenge (either gets challenged or is initiating a challenge), the agent (as he is the honest actor) can optimistically call prove, initiating the 7 day finality window.
This is only a tradeoff as the L1 transaction imposes cost to the agent which he might not get back (the malicious actor could stop engaging in the challenge, thus the agent wins anyways and does not need to call L1 resolution).
This optimistically proving in case of a challenge is actually very similar to why agents send invalidFill transactions on L2. They never might need it, but it's necessary to have the message ready in 7 days from now.
However, the claimStake does probably not cover the cost of the prove transaction (300k gas), so the agent enters into a financial risk. This fact needs to be removed in the future but it is a viable short term solution, without needing the contract to change.
What needs to be done
- The relayer needs to be able to call the prove transaction. It should only exit successfully if the transaction was successful. This is the only signal to the agent that the 7 day window started and it does not need to call the relayer again for a prove.
- Update Optimism SDK to support Bedrock in the relayer
- The agent needs to call the relayer to prove a message whenever and as soon as he is part of a challenge game where the target chain is Optimism. The agent needs to prove either a fill transaction or an invalid fill.
- We need to do research and come up with a proper solution which regains incentive alignment on all actors in the system.
- Test it on Goerli and Mainnet
GabrielBuragev
andfletcher
Rationale
OP progresses and migrates their network to the new Bedrock version.
We need to make sure our system provides the same service as before. Things to look into:
STATE_ROOT_NOT_PUBLISHEDerror in the OP e2e test related to the Bedrock version on Goerli?finality_periodchanged on OP Bedrock? Do we use the samefinality_periodin our contracts as it is defined in their system?