Pop Chunhapanya
Pop Chunhapanya
I think the purpose of IWANT is to make sure that the message is propagated across all the nodes. Without IWANT, some nodes will not receive the message, if the...
Referencing https://github.com/ethereum/consensus-specs/pull/3900#issuecomment-2383952335 and https://github.com/ethereum/consensus-specs/pull/3787 which have kind of similar ideas.
> > An alternative to not sending any IDONTWANT messages would be to only send IDONTWANTs for messages larger than a specific value (e.g., 1KB). > > nim-libp2p already does...
@leolara please reopen it
I would like to oppose **Approach 2**, as it completely eliminates anonymity of validators. Some people said there is already no anonymity in the p2p network, so we don't have...
> and (inevitably) increase block propagation time Currently we use the block propagation time as a metric to see if the network is doing well or not. See [1]. If...
> I don't know how much of a bandwidth/latency gain we could make if we sacrificed pseudo-anonymity. If its >95%, perhaps it's worth it? Maybe it doesn't matter here, but...
> * lower bound: we have a symmetric D-regular graph with N nodes, which means we have N * D / 2 edges. A message will pass on each link...
> Validators don't subscribe to subnets unless they are aggregators. Yeah, I wasn't aware that validators don't subscribe to subnets. That kind of makes sense because finding and connecting to...
> I'd create a node on the subnets, set my mesh degree to 20,000 It's also good to note that that requires like 2,000 times more bandwidth than normal full...