switch to a single list of historical batch summaries

specs/altair/beacon-chain.md

@@ -52,7 +52,7 @@
     - [Rewards and penalties](#rewards-and-penalties)
     - [Slashings](#slashings)
     - [Participation flags updates](#participation-flags-updates)
-    - [Historical roots updates](#historical-roots-updates)
+    - [Historical batches updates](#historical-batches-updates)
     - [Sync committee updates](#sync-committee-updates)
 - [Initialize state for pure Altair testnets and test vectors](#initialize-state-for-pure-altair-testnets-and-test-vectors)
 
@@ -175,8 +175,7 @@ class BeaconState(Container):
     latest_block_header: BeaconBlockHeader
     block_roots: Vector[Root, SLOTS_PER_HISTORICAL_ROOT]
     state_roots: Vector[Root, SLOTS_PER_HISTORICAL_ROOT]
-    historical_block_roots: List[Root, HISTORICAL_ROOTS_LIMIT]
+    historical_batches: List[HistoricalBatchSummary, HISTORICAL_ROOTS_LIMIT]
-    historical_state_roots: List[Root, HISTORICAL_ROOTS_LIMIT]
     # Eth1
     eth1_data: Eth1Data
     eth1_data_votes: List[Eth1Data, EPOCHS_PER_ETH1_VOTING_PERIOD * SLOTS_PER_EPOCH]
@@ -205,6 +204,15 @@ class BeaconState(Container):
 
 ### New containers
 
+
+```python
+# The tree roots that make up a phase0 HistoricalBatch making its layout
+# hash_tree_root-compatible.
+class HistoricalBatchSummary(Container):
+    block_batch_root: Root
+    state_batch_root: Root
+```
+
 #### `SyncAggregate`
 
 ```python
@@ -600,7 +608,7 @@ def process_epoch(state: BeaconState) -> None:
     process_effective_balance_updates(state)
     process_slashings_reset(state)
     process_randao_mixes_reset(state)
-    process_historical_roots_update(state)
+    process_historical_batches_update(state)
     process_participation_flag_updates(state)  # [New in Altair]
     process_sync_committee_updates(state)  # [New in Altair]
 ```
@@ -689,17 +697,19 @@ def process_participation_flag_updates(state: BeaconState) -> None:
     state.current_epoch_participation = [ParticipationFlags(0b0000_0000) for _ in range(len(state.validators))]
 ```
 
-#### Historical roots updates
+#### Historical batches updates
 
-*Note*: The function `process_historical_roots_update` changes definition compared to phase0.
+*Note*: The function `process_historical_batches_update` replaces `process_historical_roots_update` in phase0.
 
 ```python
-def process_historical_roots_update(state: BeaconState) -> None:
+def process_historical_batches_update(state: BeaconState) -> None:
     # Set historical block root accumulator
     next_epoch = Epoch(get_current_epoch(state) + 1)
     if next_epoch % (SLOTS_PER_HISTORICAL_ROOT // SLOTS_PER_EPOCH) == 0:
-        state.historical_block_roots.append(hash_tree_root(state.block_roots))
+        historical_batch = HistoricalBatchSummary(
-        state.historical_state_roots.append(hash_tree_root(state.state_roots))
+            block_batch_root=hash_tree_root(state.block_roots),
+            state_batch_root=hash_tree_root(state.state_roots))
+        state.historical_batches.append(historical_batch)
 ```
 
 #### Sync committee updates

specs/altair/fork.md

@@ -58,20 +58,21 @@ def translate_participation(state: BeaconState, pending_attestations: Sequence[p
             for flag_index in participation_flag_indices:
                 epoch_participation[index] = add_flag(epoch_participation[index], flag_index)
 
-def backfill_historical_roots(state: BeaconState, pre: BeaconState) -> None:
+def load_historical_batches(state: BeaconState, pre: BeaconState) -> List[HistoricalBatchSummary, HISTORICAL_ROOTS_LIMIT]:
-    # Clients need to recreate `historical_block_roots` and
+    # Clients need to recalculate the historical batches from genesis and return
-    # `historical_state_roots` from their history. A large part can be
+    # them here with the roots of the blocks and states separated. It's assumed
-    # precalculated and verified against historical_roots in state.
+    # most of these are pre-calculated and the rest are stored on-the-fly by
-    # Before the fork, clients should start collecting the part that has not
+    # altair-compatible clients leading up to the fork. All in all, this should
-    # been pre-calculated.
+    # be around 10kb of hashes.
-    #
-    # The calclulated split historical roots can be verified against the existing
-    # historical_roots field up to the fork.
-
 
+    return []
 
 def upgrade_to_altair(pre: phase0.BeaconState) -> BeaconState:
     epoch = phase0.get_current_epoch(pre)
+    historical_batches = load_historical_batches()
+
+    # Tree hash should match since HistoricalBatchSummary is an intermediate step in calculating the phase0 historical root entry
+    assert hash_tree_root(historical_batches) == hash_tree_root(pre.historical_roots)
     post = BeaconState(
         # Versioning
         genesis_time=pre.genesis_time,
@@ -86,8 +87,7 @@ def upgrade_to_altair(pre: phase0.BeaconState) -> BeaconState:
         latest_block_header=pre.latest_block_header,
         block_roots=pre.block_roots,
         state_roots=pre.state_roots,
-        historical_block_roots=[],
+        historical_batches=historical_batches,
-        historical_state_roots=[],
         # Eth1
         eth1_data=pre.eth1_data,
         eth1_data_votes=pre.eth1_data_votes,
@@ -113,8 +113,6 @@ def upgrade_to_altair(pre: phase0.BeaconState) -> BeaconState:
     # Fill in previous epoch participation from the pre state's pending attestations
     translate_participation(post, pre.previous_epoch_attestations)
 
-    backfill_historical_roots(post, pre)
-
     # Fill in sync committees
     # Note: A duplicate committee is assigned for the current and next committee at the fork boundary
     post.current_sync_committee = get_next_sync_committee(post)