this.
“"The local states have custom validation/consensus criteria and finality. These local states converge into the global state with its own, typically longer finality window of 20 seconds or so."
This basically means that atomic composability between state channels is not possible without using workarounds like state/liquidity locking (with bad UX).
E.g. you have a dex pool A in state channel 1 and a pool B in state channel 2, if you want to do a cross-state channel swap with atomic composability you need to lock the pool while you are waiting for the response from the other state channel which kind of reduces your throughput to 1 trade per 20+ seconds.
(That's why Elrond's dex Maiar is running on one shard only and not splitted across many shards for better scalability)
Allowing different finality and consensus for different state channels just additionally increases your problem.
Also what do they mean exactly with "atomic composability at the Layer 0".
Or do they have a some documentation which explains how this could work?
But sure, just keep me updated when they have their documentation online and I take a look.
Right now, I would say these are quite big issues for supporting atomic composability cross-state channel:
- custom finality and consensus for different state channels
- long converging time of local state to global state of 20 seconds”
@criptix for his expertise on lattice
“e.g. you imagine the following transactions sending token A from state channel 1 to state channel 2 sending these tokens from state channel 2 to state channel 3 sending these tokens now to state channel 3 and that could either accept or deny the tokens Now if the last part fails and everything the transaction is aborted you could say you have "atomic" composability. Problem is that this is fully asynchronous and you don't have dependencies like with my dex example”
also what is meant here with the last sentence "- long converging time of local state to global state of 20 seconds”? 1.) what does converging time in this context mean? 2.) is that a fact with the 20 seconds between local and global state? 3.) Is that not a big disadvantage compared to the current iot?
can someone reply to this? maybe even you @SethVi
Look at State channels as their own independent dag networks which have their own finality. Then all of them converge into the Hypergraph network itself that reflect total finality of the entire network itself
Ok so as far as I understand your answer that means that the convergence is all state channels/independent dag network reflecting to the hypgergraph network itself and that needs 20 seconds right? But then when will those convergences happen if we have 1000 state channels for example, will we have every 20 second an update on the current hypergraph network itself? (of course just 20sec intervals meaning that transaction which started after the last intervall will be shown in the next one)
Обсуждают сегодня