@p >Then one wonders why it has been completely unsuccessful at dislodging IPv4 for...three decades.
For 20 years IPv6 wasn't supported in big routers due to software patents - as companies clearly didn't want to be faced with paying millions in patent licensing.

Now all patents of the base IPv6 spec have expired, big routers now support IPv6.

After that, the problem seems to be "everyone keeps using IPv4 despite its brokenness, so we'll keep using IPv4".

>Of course, "Just a software change!" presumes changeable software, it presumes that changing is zero-cost, and it ignores infrastructure and coordination.
Changing it has already been done and support is there in infrastructure, the problem is few are willing to coordinate on anything at all.

>No, they always wanted a complete overhaul before the address exhaustion was a factor.
They accurately predicted address exhaustion well in advance, so they decided to start on the fix right away, to ease the transition, rather than it being excessively abrupt.

>The Second System was always going to fix *all* of the problems of the first one. It's the real silver bullet this time, guys.
IPv6 doesn't fix all networking problems, but it does fix all the problems that IPv4 has.

>see in RFC 1627 that they argue *against* private subnets partially because delaying address exhaustion will delay the adoption of the new network protocol that will Fix Everything.
It's entirely valid to argue against the brokenness of private subnets noting that it will delay the adoption of the network protocol that fixes the address exhaustion problem once and for all.

>If it's adopted at the scale of IPv4, we'll know if it works better.
IPv6 has been adopted at scale in the backend of all modern tracking device networks (as newer mobile protocols require IPv6) and it works just fine - even better than IPv4 as such spying companies found out (of course those companies proceeded to setup only IPv4 gateways for internet access and then route the IPv4 packets over IPv6 to reach the internet gateway).

Google for their botnet activities have noted IPv6 works just fine at huge scale with the right happy eyes configuration.

>DHCP seems to work fine on this network.
Yes, DHCP seems to work fine until it doesn't and then good luck troubleshooting.

>Meanwhile, unless I pass "-4" to curl/wget, I get to sit and watch them time out.
Your ISP's uplink to the internet is very broken - I would suggest you either complain to your ISP and demand that they fix your internet connection or you go to someone else (here's hoping that there isn't a monopoly that stops you from doing that), or seeing if you can fix your routers configuration.

The same thing would happen if your ISP only supported IPv6, but left a broken IPv4 route - your problem really has nothing to do with IPv6.

>"It's a shitty notation" is not answered by "Well, just copy and paste."
Your complaint seems to be based on how typing out 128 bits of information is inconvenient, so using a computer to send that exact information where it needs to go without error-prone manual entering it is the answer.

The notation scheme is fine in my experience and the abbreviation standard is okay, but of course no notation is ever perfect - do let me know if you have a better encoding scheme.

>* 64 bits would have been fine.
No it wasn't - if a 64 bit addressing scheme was used, address exhaustion would eventually happen again.

A 64 bit addressing scheme is just an incompatible with a 32 bit addressing scheme as a 128 bit addressing scheme is.

If I read correct information, IPv4 was drafted to use 128 bit addresses originally, but 32 bits were decided on as that was easier to implement on the current hardware, plus it was just for a testnet.

>* IPv4 will never go away. We are still emulating the Teletype Model 33-ASR and people are still linking against to speak VT-100 escape codes: that's way easier to drop than a network protocol.
The Teletype Model 33-ASR has indeed been dropped - the escape codes are now being emulated in software.

Internet backbone routers that only support IPv4 have been dropped as well - they now route IPv6 just as well.

Some ISP's are even offering IPv6-only by default and normies that only visit normie sites like facebook or google don't even notice.

>* IPv6 gives me nothing I want. No carrot, and the stick has still not materialized.
You don't want nice things, so IPv6 clearly doesn't give what you want.

The carrot is end to end connectivity and the stick is how the consequences of IPv4 address exhaustion will progressively get worse no matter how many levels of CGNAT are thrown at it.

>* I will care about IPv6 when I have to. I do not have to, and will not have to for years to come.
The idea is really you don't have to care about IPv6 - it's just implemented and it just works, but some ISPs seem to have made the intentional decision to prevent that from happening.

>* It is not desirable that every system be globally routeable, which is one of the explicit goals of IPv6.
Either something is connected to the internet, or it's not.

IPv4 is globally routable, no matter how many layers of NAT - as otherwise you wouldn't be able to send and receive packets from the global internet - it's doubethink to think that an internet connected IPv4 network isn't connected to the internet because of NAT.

The explicit goal of IPv6 is to make it so that if a network is part of the internet, you can send packets to all other computers on the internet without having to resort to nasty workarounds like NAT holepunching (accepting the packets is left up to the receiving computers, which is what should happen).

If you don't want your system to be globally routable, you need to not connect it to the internet.