Open sebastianst opened 1 month ago
@sebastianst do you think if we implement https://github.com/ethereum-optimism/optimism/issues/12124 it might just handle this case?
Answering my own question -- experimenting with #12430 I realise that that feature will have some inherent delay or lag (at least a few L1 block times). This might be too slow for Holocene activation (we know that situation is coming so we can probably respond to it quicker).
When Holocene activates, the derivation pipeline state will be mostly reset, per spec. So the batcher needs to be aware of this rule in the derivation pipeline, namely: If the batcher has only partially submitted a channel pre-Holocene and then the L1 timestamp passes the Holocene activation timestamp, it has to submit the first frame(s) of the channel again, because they would be dropped in the frame queue/channel bank at Holocene activation.