+ // TODO: We need to consider HTLCs which weren't included in latest local commitment
+ // transaction (or in any of the latest two local commitment transactions). This probably
+ // needs to use the same logic as the revoked-tx-announe logic - checking the last two
+ // remote commitment transactions. This probably has implications for what data we need to
+ // store in local commitment transactions.
+ // TODO: We need to consider HTLCs which were below dust threshold here - while they don't
+ // strictly imply that we need to fail the channel, we need to go ahead and fail them back
+ // to the source, and if we don't fail the channel we will have to ensure that the next
+ // updates that peer sends us are update_fails, failing the channel if not. It's probably
+ // easier to just fail the channel as this case should be rare enough anyway.