Comment by pif
5 days ago
I think 30 years should be much more than enough to realise the idiocy of proposing a non-backward-compatible standard to the general public.
5 days ago
I think 30 years should be much more than enough to realise the idiocy of proposing a non-backward-compatible standard to the general public.
We replaced VHS with DVDs. It took 42 years before we gave up on VHS. DVDs have been around for 29 years but were mostly replaced with BDs before disappearing off the shelves in favor of streaming.
We replaced records with tapes, tapes with more tapes, and more tapes with CDs before they, too, disappeared from the shelves in favor of streaming. Except that some stalwarts have successfully resurrected vinyl.
We replaced AM with FM, and analog radio with digital radio, then streaming. We replaced broadcast analog TV with digital, then cable and satelite, then streaming. Mostly.
None of these changes were backwards compatible, and all of them were meant for the general public. They took a while. They were successful.
The quality jump from vhs to dvd was massive. In comparison v6 doesn't offer much above v4
Yes, I've never played a DVD or CD on my Bluray player. That just didn't works.
Anyone who bought DVD player immediately had the benefits of better quality. The same applies to all other examples.
The problem with IPv6 is that you don't get benefits. If the designed protocol needs an equivalent of big bang, it's doomed. ASCII->UTF8 didn't need big bang. x86 to Itanium needed big bang.
You'd think it would be long enough for people to realize that v6 is backwards compatible! Yet no, here we are, constantly dealing with people making the same damn claim that it isn't every single time a v6 story is posted.
v6 is about as backwards compatible with v4 as it's possible to be. If you have a way to make it more backwards compatible then I'd love to hear it, but when I ask this all I ever get are things that don't work, or things that v6 already does.
No, it's not. If I have an ipv6 network, an ipv4 address is invalid. It's that simple.
It's not that simple at all. For one thing, having a v6 network doesn't mean you can't have a v4 network. You can run v4 in exactly the same way you currently do, with exactly the same software, and it'll work no worse than it already does.
But for another, the v4 space is available as a subset of the v6 space:
That's from a machine on a network with no v4, and it works fine. I can reach v4-only sites from it too. I could even do this using v4 addresses if I wanted, but if I showed you the output from that you'd just claim I was using v4.
2 replies →
No, it’s not—IPv6 networks are totally capable of providing IPv4 as a service. SIIT-DC, 464XLAT, MAP-T
It's often impossible to make backwards-compatible changes to a format which wasn't designed to allow for future changes and which is designed to be as space-efficient as possible.
That doesn't mean that the limits of the old design won't hit anyway and force a switch off it.
IPv4 allows future changes. There are some reserved bits in the header that could change a big part of it.
v4 supports extension headers and over a thousand bytes of arbitrary payload so if the only thing you needed was a couple of bits in the packet, there was never any issue with finding them.
The problem is that you can't use those bits to expand v4's address space, without taking all of the same steps v6 needed to do. v4 has no mechanism to get v4 hosts to understand extra address bits, wherever you put them.
Oh, that and the fact that IP addresses are stored in many more places than just the v4 packet header. Consider DNS, DHCP, ARP, gethostbyname(), struct sockaddr_in, databases using VARCHAR(15), etc etc etc. The packet header is only a tiny part of the story.
The problem is that IPv4 has no provisions to be forward-compatible with anything with a larger address space. Thus whatever replacement you can think of will have the same problems as IPv6.