@jeff @Moon @admin @graf @john_rando @mint @parker @sjw @verita84

> i think given the right amount of technical rigor ipv6 is worth it,

The deal is that there's nothing really compelling. It doesn't do anything that I want to do except provide more addresses, and it fucks that up with way too many addresses. What's the payoff for the effort? The internet isn't going to stop being a complete mess, and I don't think IPv6 meaningfully changes anything for anyone besides people that hack routers and BGP advertisements, aside from obliging people to cope with a horrible notation. (And one that makes previously easy things into things that most people cannot do. I can remember 32 bits for long enough to look at it in one place and type it in another. How many people can do this with 128 bits?)

> software broken by ipv6 is a software bug,

I don't think anything is broken by IPv6, unless you mean software that does not support IPv6.

> someone needs to make an LD_PRELOAD hack to auto promote v4 to the v6 translation layer.

This is not only a bad idea, it's the worst way to do it. You do it with the LAN's router. Good luck LD_PRELOAD'ing firmware on a machine that's welded shut. Sure, they should have gone with an open solution, but they didn't, and no amount of "but the checksum calculations!" is going to be worth shutting down operations.

The sheer amount of unpatchable, deployed hardware that does not understand IPv6 is more massive than all of the industrial systems that still use DOS, which is also never going away. You can say it's a bug, but a thing that was working as designed and that still works as designed is not buggy. In this case, you don't like the spec it was designed for. That's fine, but trying to force the new protocol, even if it were a better protocol, never works.