"The transition has sort of already happened," he says, over IPv4.
Seriously, unless you want to be posting to HN via IPv4 in 2050, there is clearly something major left undone. Don't call it a transition if you think that concept is undone, but it's not yet functionally possible to be on the Internet with no IPv4 stack or (possibly NATted) address, and I don't see that becoming the case anytime soon.
See http://cr.yp.to/djbdns/ipv6mess.html for a description of the problem that I mostly agree with. The depressing thing is that those complaints are no less valid a decade later.
Only because HN is IPv4-only. I'm sitting in a remote office on my laptop this morning, and it has an IPv6 address. When I visit google.com, it's over IPv6.
Why is HN IPv4-only? And why should they bother to spend any effort fixing it? Given that, e.g., techcrunch.com and github.com and everything else linked from HN is IPv4-only, you can't usefully use Hacker News from an IPv6-only connection, so why should HN care?
You're expecting altruism or geekiness on the part of every single website ever -- it's fair to expect that from Google, but not from everyone. The transition may or may not be happen_ing_, but it will never finish.
No I'm not, I'm expecting anti-altruism from low-cost ISPs. (Name one consumer ISP that you think is altruistic.) Then, to capture that market, servers will start deploying IPv6.
Today, IPv6 is billed as a premium feature. Tomorrow, IPv4 will be the premium feature.
Wait, you expect any low-cost ISPs to offer IPv6-only, and have any customers at all? And that IPv4-only sites will say "gee, better support those IPv6-only users" instead of saying "get a better ISP" (or more likely "We have no idea what's wrong, but it works from Comcast, complain to your ISP")?
I don't think you understand how the Internet works. Or business. There is no foreseeable point at which it makes sense to bill an "Internet" service that's IPv6-only. That's doable _after_ the transition, but expecting that to happen now is not a transition.
I feel like a lot of the issues raised in that article aren't as relevant anymore. The issue of IPv6 only clients communicating over IPv4 can be solved via tunneling in both directions. Reading over that article again, it really does seem like he's complaining about the transition not the protocol. The solution to what he's complaining about is being solved by a slow and gradual approach that's taken place for well over a decade. The details have been worked out as time has gone on.
The biggest issue will not be getting everybody onto IPv6. That will happen at some point, however distant. The bigger issue is that clients slow to transition to IPv6, for whatever reason, will experience IPv4 service degradation as more and more of the internet switches to an IPv6 only stack.
Yes, he's complaining about the transition, not the protocol. One perfectly acceptable transition method, from a technical viewpoint (though probably not a political one), is the one that got us on IPv4 in the first place: a declaration that the Internet's core routers will stop routing the old protocol come January 1. Another transition method would be to write a protocol that doesn't need such tactics to get people to switch to.
If it's been "being solved" over the last ten years, neither you or I will see the end of IPv4 on the public Internet within our lifetimes. Or, if we do, it's because something bigger and better will come along, obsoleting every version of IP. The debate over whether to switch to IPv6 will sound as silly as a debate about Morse code in the 1960s.
The future will come up with their protocols. IPv6, or maybe something other than IPv6, needs to be a protocol we can start using now (and not just because it's cool, but because it's useful) and not feel like it's wasted if it becomes irrelevant in 2025.
Seriously, unless you want to be posting to HN via IPv4 in 2050, there is clearly something major left undone. Don't call it a transition if you think that concept is undone, but it's not yet functionally possible to be on the Internet with no IPv4 stack or (possibly NATted) address, and I don't see that becoming the case anytime soon.
See http://cr.yp.to/djbdns/ipv6mess.html for a description of the problem that I mostly agree with. The depressing thing is that those complaints are no less valid a decade later.