The issue here is that there are two different ways to assign addresses in IPv6: SLAAC and DHCPv6. Apple only wants to support SLAAC; they think it's a waste of their time to have two different protocols to do the same thing. Unfortunately, a lot of ISPs want to only support DHCPv6, because it (easily) supports an audit trail and SLAAC doesn't.
I predict that things will turn out fine because people will converge on speaking SLAAC between OS X and the router and DHCPv6-PD between the router and the ISP. But you have to have a router.
Depending on DHCP to audit trail is flawed, because it is easy to spoof addresses.
Unless, of course, you're restricting traffic to prevent spoofing: in which case you could be doing the same for addresses assigned by SLAAC just as easily.
SLAAC also has a lot to offer for faster configuration of mobile devices roaming between networks.
Apparently OS X now supports supplying DNS servers through SLAAC. Previously this wasn't supported, so you'd still need a DHCP(v4) server to fully support a client.
> Apparently OS X now supports supplying DNS servers through SLAAC
i thought this was done via router-advertisements only. slaac is used to configure
v6 host addresses, and once that is done, other network parameters e.g. rdnss etc. can be obtained either via RA or DHCPv6 (the 'O' flag in RA messages)
To Apple's credit, they are one of the few vendors of consumer-grade routers who made it trivial to set up IPv6 (Airport Extreme/Express), including tunneling.
On what basis should Apple be forced to provide a DHCPv6 client as part of their OS? They have 10% of the market, and the number of IPv6 users I imagine is going to be a niche within their niche.
I appreciate that DHCPv6 support may be useful for some apple customers, but in the grand scheme of things it's probably in a minority, perhaps best served by the app store.
I also like the idea of references to 'religious nuts' being tied to apple users, not all but specifically of the fanboy type.
In short, I find the statement poetic. The value it adds is humour. You might not agree as humour is subjective, but it works for me.
No, on the contrary. What I'm saying is that IPv6 doesn't have enough traction, there's a finite number of developers and there are features that will be used by more people that should take more priority.
To put it another way, a native IPv6 client isn't necessarily as innovative as some of the other pending features. Not having DHCPv6 native isn't the end of the world and is a minor annoyance for a few. Putting a team on implementing DHCPv6 for a small subset of users takes people away from things like bugfixes. DHCPv6 is not trivial.
If Apple found a feature that relied upon IPv6 then I imagine a DHCPv6 client would be a priority. As it is, in the meantime you'll have to settle for things like http://klub.com.pl/dhcpv6/ - which at least works on leopard but you'll have to compile it yourself and it's not complete.
They don't have to yet. Really, IPv6 is not going to be an issue this year at all. That ends up just being a patch Apple can come up with later. I don't see the big deal here.
Right. The Apple developers have been working very hard on 10.7, and it's kind of sad to see a story focusing on some vague subject (which, as you say, can be applied as a patch anytime later when it becomes more relevant) that is lacking. In my opinion, it's better to offer constructive criticism on the new stuff that they've been working on for this release.
Stateless autoconfig is, but stateless is bad for ISPs and network administrators because it means anyone will be able to get an IP just by plugging their machine into the network. With only 3.40282367 × 10^38 IPs available, you can see why they would want to ration them.
At current population growth, we will run out of addresses—if we give every human one IPv6 address—in only ~6000 years, or—if everybody should get 1000 addresses, for all those internet-enabled toasters and such—in ~5500 years. Obviously we already need to plan ahead and carrier-NAT IPv6.
according to node-requirements rfc-4294, dhcpv6 is infact optinal. however, for an enterprise network, the fact that any v6 capable device can get a legit address is probably not such a good idea (it can be argued the other way also ofcourse).
Yes, stateless autoconfiguration in IPv6 is supported by OS X. However, for stateful autoconf, you still need (a newer version of) DHCP, which OS X doesn't support yet.
I predict that things will turn out fine because people will converge on speaking SLAAC between OS X and the router and DHCPv6-PD between the router and the ISP. But you have to have a router.