Potential bug in state transition

[ChihChengLiang]

What is wrong?

trinity/eth2/beacon/state_machines/forks/serenity/state_transitions.py

Lines 42 to 62 in af4bc6b

	def apply_state_transition(self,
	state: BeaconState,
	block: BaseBeaconBlock,
	check_proposer_signature: bool=True) -> BeaconState:
	if state.slot >= block.slot:
	return state
	for _ in range(state.slot, block.slot):
	state = self.cache_state(state)
	if (state.slot + 1) % self.config.SLOTS_PER_EPOCH == 0:
	state = self.per_epoch_transition(state)
	state = self.per_slot_transition(state)
	if state.slot == block.slot:
	state = self.per_block_transition(state, block, check_proposer_signature)
	break
	else:
	raise Exception(
	f"Invariant: state.slot ({state.slot}) should be less "
	f"than block.slot ({block.slot}) so that state transition terminates"
	)
	return state

There should be a per_block_transition on state.slot == slot for the case that the state is a pre-state of a block.

How can it be fixed

The spec now has a python state transition function specified, maybe can be fixed in #714.

[ralexstokes]

@ChihChengLiang is this not what is happening on L55?

could you elaborate what you think is wrong or post a code snippet to illustrate the bug?

happy to update this code but i don't see the problem at the moment

[ChihChengLiang]

This is no longer an issue after the spec update.

The issue was if a state is a pre-state of a block, then the state and the block has the same slot. So everything in the loop is noop, including L55.

In [1]: for _ in range(5, 5):
   ...:     print(_)
   ...:     
<nothing>