On: Making peer-to-peer multiplayer seamless with Godo...

[jesprenj]

So they don't need a server, but they need a server at the same time (:

One could abuse BitTorrent's existing DHT infrastructure (:

[imtringued → jesprenj]

What we need is widespread IPv6 support. Portforwarding isn't the problem. The terrible UX and the existence of CGNAT (carrier grade NAT) is.

[jesprenj → imtringued]

While I agree with you, having a globaly routable address doesn't help you if you can't accept incoming tcp/udp connections.

My ISP blocks such connections on LTE mobile network. Maybe so you don't pay a lot of money for traffic if people DoS you? It doesn't make sense, since ICMP ping packets are allowed.

Well, it's either my ISP or the android phone I have. Does android have any firewalls enabled by default?

[Zetaphor → imtringued]

As someone who only understands the benefit of IPv6 to be the much larger address pool, can you elaborate on how it would make this situation simpler?

[chainingsolid → Zetaphor]

https://tailscale.com/blog/how-nat-traversal-works/ This was posted to HN recently and has the answer you asked for better written up the I can write here.

[amarshall → Zetaphor]

The problem is NAT as it makes it impossible to directly address, and NAT doesn’t exist in IPv6 (because NAT is primarily needed to workaround limited public addresses).

[grumpy_coder → imtringued]

IPv6 support is pretty 'widespread' at this point. >50% last I did a quick test

You unfortunately need ubiquitus IPv6 support to actually rely on IPv6, which isn't going to happen anytime soon.

So you really need to consider the benefits of dual stack v4 and v6 simultaneously within the same group of players. Might be useful, but it adds a whole heap of complexity.

[p1mrx → imtringued]

CGNAT is not a problem. It is a necessity because there are more humans than IPv4 addresses.

The problem is deploying CGNAT without also providing IPv6 as an alternative.