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
{{ message }}
This repository has been archived by the owner on Jan 29, 2024. It is now read-only.
The reason for this is that we assume that all subnet will be spawned with epoch 0, and thus, new bottom-up checkpoints must be submitted from the subnet's genesis. This is true for now, but once we support the docking of IPC-compatible subnets, or subnets changing parents, we will need to set the genesis epoch to the first epoch where the network docked to the new parent instead of 0. This way, the subnet actor doesn't require the commitment from every checkpoint from genesis to be able to execute from the moment it docked to the new parent.
Having genesis_epoch != 0 will only be needed when we allow a subnet to become a child from an IPC parent when it is already mining blocks to avoid having to submit every checkpoint from zero. This is a consequence of the new design for checkpoint voting, where we don't allow gaps between checkpoints to ensure that all checkpoints (and the underlying cross-net messages included) are executed.
The text was updated successfully, but these errors were encountered:
Currently we use
genesis_epoch = 0
by default in all subnets:ipc-actors/subnet-actor/src/lib.rs
Line 89 in 5ef2203
The reason for this is that we assume that all subnet will be spawned with epoch 0, and thus, new bottom-up checkpoints must be submitted from the subnet's genesis. This is true for now, but once we support the docking of IPC-compatible subnets, or subnets changing parents, we will need to set the genesis epoch to the first epoch where the network docked to the new parent instead of 0. This way, the subnet actor doesn't require the commitment from every checkpoint from genesis to be able to execute from the moment it docked to the new parent.
Having
genesis_epoch != 0
will only be needed when we allow a subnet to become a child from an IPC parent when it is already mining blocks to avoid having to submit every checkpoint from zero. This is a consequence of the new design for checkpoint voting, where we don't allow gaps between checkpoints to ensure that all checkpoints (and the underlying cross-net messages included) are executed.The text was updated successfully, but these errors were encountered: