-
Notifications
You must be signed in to change notification settings - Fork 69
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
chore: Remove reveal
roundtrip
#2942
Comments
Related to #2939 which mentions that calldata is not useful anywhere at all anymore. |
@harisang I am letting you know I will start implemented this. This can potentially affect co-located solvers. |
@fleupold where can I find the documentation to alter this docs: https://docs.cow.fi/cow-protocol/tutorials/arbitrate/autopilot ? after this is merge, this docs will be out of date. |
@fleupold one question about this sentence
Do we need to do that in the backend in a background job, or will it be solver's team responsibility? |
The source code for the docs is here: https://github.com/cowprotocol/docs
If we want to keep the calldata on this endpoint, then yes it would have to be added in |
The solver team currently does not (edit: forgot that word) use call data from the competition endpoint for anything (that I know of). |
Yeap, as @fhenneke said above, we are fine with fully dropping calldata and not making the available at the competition endpoint. |
# Description Currently, we have no use for reveal and we are calling settle unconditionally after it. We are storing the internalised and uninternalised calldata that is returned in the database. However, since slippage accounting moved towards counting internal buffer trades as slippage we have no more use for the "uninternalised" calldata. The final calldata can also be recovered from the settled transaction, and doesn't matter in the case settlement failed. # Changes - Remove `/reveal` roundtrip - Remove internalised and uninternalised calldata from the competition/solution objects ## How to test 1. Acceptance tests ## Related Issues Fixes #2942
# Description Currently, we have no use for reveal and we are calling settle unconditionally after it. We are storing the internalised and uninternalised calldata that is returned in the database. However, since slippage accounting moved towards counting internal buffer trades as slippage we have no more use for the "uninternalised" calldata. The final calldata can also be recovered from the settled transaction, and doesn't matter in the case settlement failed. # Changes - Remove `/reveal` roundtrip - Remove internalised and uninternalised calldata from the competition/solution objects ## How to test 1. Acceptance tests ## Related Issues Fixes #2942
Background
The current runloop first asks the winning solver to "reveal" their calldata before telling them to submit their transaction. The result is unused and can add additional latency on the critical path (current timeout is 10s I believe). It should be removed
Details
Reveal
was originally introduced because we thought about doing some checks on the proposed solution and potentially signing it off, however as things panned out we now have no use for this information and are callingsettle
unconditionally after reveal. We are storing the internalised and uninternalised calldata that is returned in the database. However, since slippage accounting moved towards counting internal buffer trades as slippage we have no more use for the "uninternalised" calldata. The final calldata can also be recovered from the settled transaction, and doesn't matter in the case settlement failed.We might want to keep
reveal
for the shadow competition, where we don't ask solvers to settle and thus could use this information for counterfactual analysis of the provided solution (e.g. running the circuit breaker on it to make sure a solver is not violating any rules)Acceptance criteria
No longer call
reveal
in the "real" (ie non-shadow) run-loop. This change needs to be communicated with co-located solvers to make sure they don't rely on this being called.The text was updated successfully, but these errors were encountered: