Closed StefanBratanov closed 3 months ago
I think we would need to group IHAVE
s in the resulting message produced by GossipRpcPartsQueue
else we might get significant message size overhead
I think we would need to group
IHAVE
s in the resulting message produced byGossipRpcPartsQueue
else we might get significant message size overhead
Isn't that what the GossipRouter
does in emitGossip
on a heartbeat and only then enqueues it?
val shuffledMessageIds = ids.shuffled(random).take(params.maxIHaveLength)
I suppose we could be more efficient when flushing the queue. Is it worth to make this improvement a separate PR?
Please check here: https://github.com/StefanBratanov/jvm-libp2p/pull/1 Fixing the case has also addressed the grouping question
Please check here: StefanBratanov#1 Fixing the case has also addressed the grouping question
Thanks. I agree with the change. I have missed that part. Thanks.
Added another minor improvement
fixes https://github.com/libp2p/jvm-libp2p/issues/361