You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Because of time pressure for nv23 we were forced into a corner and could not completely remove all traces of FIP 0084. We had to keep around a version of CSPV renamed to InteranlSectorSetupForPreseal: #1553
This whole PR should be reverted. We should not be shipping this endpoint to the miner actor in the production bundle.
The reason we need it is that lotus-seed, which is used in devnets, butterfly and rarely calibration reset, needs a way to mock out miner state. The correct way to do this is to use a sector proving method, however this requires either sealed sectors with proofs which is not feasible for large networks or the ability to mock out syscalls.
The correct approach is to mock out syscalls. However this is technically challenging since the FVM was introduced. When @aarshkshah1992 and I looked at this last week it appears that this will require significant engineering work on the FVM and its ffi to the lotus node.
The text was updated successfully, but these errors were encountered:
Because of time pressure for nv23 we were forced into a corner and could not completely remove all traces of FIP 0084. We had to keep around a version of CSPV renamed to InteranlSectorSetupForPreseal: #1553
This whole PR should be reverted. We should not be shipping this endpoint to the miner actor in the production bundle.
The reason we need it is that lotus-seed, which is used in devnets, butterfly and rarely calibration reset, needs a way to mock out miner state. The correct way to do this is to use a sector proving method, however this requires either sealed sectors with proofs which is not feasible for large networks or the ability to mock out syscalls.
The correct approach is to mock out syscalls. However this is technically challenging since the FVM was introduced. When @aarshkshah1992 and I looked at this last week it appears that this will require significant engineering work on the FVM and its ffi to the lotus node.
The text was updated successfully, but these errors were encountered: