Merge pull request #4 from hwwhww/proofread
`EPOCH_LENGTH` -> `CYCLE_LENGTH`
This commit is contained in:
commit
e712cd1f04
|
@ -378,7 +378,7 @@ Let `time_since_finality = block.slot_number - last_finalized_slot`, and let `B`
|
|||
|
||||
For each slot `S` in the range `last_state_recalc - CYCLE_LENGTH ... last_state_recalc - 1`:
|
||||
|
||||
* Let `total_participated_deposits` be the total balance of validators that voted for the correct hash in slot `S` (ie. the hash that actually is the hash of the block at that slot in the current chain); note that in the normal case, every validator will be in one of the `EPOCH_LENGTH` slots following the slot and so can vote for a hash in slot `S`. If `time_since_finality <= 2 * CYCLE_LENGTH`, then adjust participating and non-participating validators' balances as follows:
|
||||
* Let `total_participated_deposits` be the total balance of validators that voted for the correct hash in slot `S` (ie. the hash that actually is the hash of the block at that slot in the current chain); note that in the normal case, every validator will be in one of the `CYCLE_LENGTH` slots following the slot and so can vote for a hash in slot `S`. If `time_since_finality <= 2 * CYCLE_LENGTH`, then adjust participating and non-participating validators' balances as follows:
|
||||
* Participating validators gain `B // reward_quotient * (2 * total_participated_deposits - total_deposits) // total_deposits` (note: this may be negative)
|
||||
* Nonparticipating validators lose `B // reward_quotient`
|
||||
* Otherwise, adjust as follows:
|
||||
|
|
Loading…
Reference in New Issue