add process_slots usage to validator guide
This commit is contained in:
parent
27c8e41b32
commit
7bb85a69ed
|
@ -238,7 +238,7 @@ A validator should create and broadcast the attestation halfway through the `slo
|
|||
First the validator should construct `attestation_data`, an [`AttestationData`](../core/0_beacon-chain.md#attestationdata) object based upon the state at the assigned slot.
|
||||
|
||||
* Let `head_block` be the result of running the fork choice during the assigned slot.
|
||||
* Let `head_state` be the state of `head_block` processed through any empty slots up to the assigned slot.
|
||||
* Let `head_state` be the state of `head_block` processed through any empty slots up to the assigned slot using `process_slots(state, slot)`.
|
||||
|
||||
##### LMD GHOST vote
|
||||
|
||||
|
@ -360,7 +360,7 @@ def is_proposer_at_slot(state: BeaconState,
|
|||
return get_beacon_proposer_index(state) == validator_index
|
||||
```
|
||||
|
||||
*Note*: To see if a validator is assigned to proposer during the slot, the validator must run an empty slot transition from the previous state to the current slot.
|
||||
*Note*: To see if a validator is assigned to proposer during the slot, the validator must run an empty slot transition from the previous state to the current slot using `process_slots(state, current_slot)`.
|
||||
|
||||
|
||||
### Lookahead
|
||||
|
|
Loading…
Reference in New Issue