Mf/generate pos leader schedules #538
Conversation
lib/pos_leader_schedule.go
Outdated
| } | ||
|
|
||
| // Retrieve top, active validators ordered by stake. | ||
| validatorEntries, err := bav.GetTopActiveValidatorsByStake(int(bav.Params.LeaderScheduleMaxNumValidators)) |
There was a problem hiding this comment.
Never use int. Always be clear about the size of the int by using int32 or int64
lib/pos_leader_schedule.go
Outdated
| } | ||
|
|
||
| // Retrieve top, active validators ordered by stake. | ||
| validatorEntries, err := bav.GetTopActiveValidatorsByStake(int(bav.Params.LeaderScheduleMaxNumValidators)) |
There was a problem hiding this comment.
TODO: Need to make sure that this function returns the top validators in a deterministic order. If it has any non-determinism in it, then nodes will have different leader schedules.
There was a problem hiding this comment.
I updated the tests so that they run GenerateLeaderSchedule() 10x times for each CurrentRandomSeedHash. It generates the same LeaderSchedule each time. I then repeat that process with 7x different CurrentRandomSeedHashes, generating different LeaderSchedules each time. With those tests, I'm fairly confident in this algorithm's determinism.
lib/pos_leader_schedule.go
Outdated
| // Take RandomUint256 modulo TotalStakeAmountNanos. | ||
| // For each ValidatorEntry: | ||
| // Skip if ValidatorPKID has already been added to the leader schedule. | ||
| // If ValidatorEntry.TotalStakeAmountNanos >= RandomUint256: |
There was a problem hiding this comment.
I think this comment is incorrect. You're tracking the SUM of the TotalStakeAmountNanos.
| } | ||
|
|
||
| // Take RandomUint256 % TotalStakeAmountNanos. | ||
| randomUint256 := uint256.NewInt().Mod(currentRandomSeedHash.ToUint256(), totalStakeAmountNanos) |
There was a problem hiding this comment.
TODO: Make sure Mod is efficient. I'm almos certain it is.
| } | ||
| } | ||
|
|
||
| return leaderSchedule, nil |
There was a problem hiding this comment.
Seems pretty clean. Good work!
No description provided.