sukun

The PR I've raised changes `host.NewStream` and `swarm.NewStream` to wait for a direct connection. Why to do this in NewStream as opposed to DialPeer. - I don't want to change...

Need to merge https://github.com/libp2p/go-libp2p/pull/2542 to close this.

I like the idea of a notification that the stream has finished sending, but that will not help with a graceful connection close. This is because QUIC doesn't specify any...

Can we move the address handling logic out and keep the peer record api as ``` type CertifiedAddrBook interface { MaybeAddCertifiedPeerRecord(envelope *record.Envelope) (updated ok) GetCertifiedRecord(p peer.ID) *record.Envelope } ```

Here is something similar that tailscale does: https://github.com/tailscale/tailscale/blob/main/tstest/natlab/natlab.go#L737

The requirement is ambiguous. We need to decide whether we want an in memory transport for testing where we can fake IP address and possibly simulate NAT or we want...

>I expect building a Transport "where we can fake IP address and possibly simulate NAT" is much harder. I think you're right. I'll change my stance to Building a transport...

@ajaygupta112 Yes! You can start here: https://github.com/libp2p/go-libp2p/pull/2531/files#diff-54f6226e2cf835d8f4f664f08e1e9a325aaa75d1bf9384c89f3c66fbd5ff8ae7 To begin with, you can enable just the Ping test for ip6.

Some racy behaviour with tcp simultaneous connect and reuseport with an Accept on the same socket as the reused port. In the event that the dial times out the sequence...

I wish I could tell you to use quic. But looks like there's a problem with simultaneous udp messages on linux :( I'm getting this error sometimes: ``` === RUN...