Comment by cyberax

> I’d say it’s a perfectly reasonable assumption. Why? In practice most mobile carriers (notably T-Mobile USA likes to discuss it) are entirely v6 only, with 464XLAT at the edges.

Carriers have their own specific challenges. Like not having enough private or even squattable IPv4 space for the voice circuits.

> If mobile carriers can do it, and support $billions of devices, it does work, and works well.

Mobile devices are actually pretty much the best case for IPv6. The phone environment is standardized and highly controlled. And yet some telecoms still don't provide end-user IPv6, even when the voice circuits use IPv6.

> No one should be using NAT with v6 except in extremely limited circumstances. NPT is not NAT, the host portion of the network and, more importantly, the port do not change with NPT.

This is really a distinction without difference. In practice, SOHO border routers have to be stateful to provide meaningful security for UDP-based protocols like SIP or QUIC.

> That is effectively how it works, they can coexist together, but it is significantly simpler to only use one.

The socket API still doesn't allow this. You have to implement Happy Eyeballs on the app level.

> I do agree with you on the ULA priority though. Was a moronic decision to place that below v4, likely again, a lack of imagination on their part.

If you want to look at politics, this was a compromise. There was a strong faction of "network fundamentalists" in the IETF that had a vision of a transparent end-to-end network with global addressing. So that a lightbulb in China could connect to a toaster in the US at will. So they resisted anything that could promote locally isolated networks.

But most of the IETF had a clue and actual operational experience, so they pushed for things like the ULA and PI space for IPv6. So we got these compromises that don't work well as a result.