Moin, > I think the number of hosts and routers on the global Internet which > have IPv6 enabled and IPv4 disabled can probably be counted on one > person's fingers and toes; see: > > > https://ungleich.ch/en-us/cms/blog/2019/02/05/list-of-ipv6-only-services/ > https://sites.ip-update.net/
Note that this is not about v6-only hosts; It is about routing, mostly; Freeing all those transfer IPs. And that actually does have some more traction going. If you want to test eBGP without v4-while-still-transporting-v4, see: https://measurement.network/services/v4less-as/ > I think that it would be useful to get v6 implementations like Linux > to ask and answer using NDP for IPv4-mapped or IPv4-compatible IPv6 > addresses, when no IPv4 stack is configured in that implementation. > Probably no sysctl would even be needed. > > Indeed, the Linux kernel should correctly *answer* any incoming IPv6 > NDP queries about such addresses, even if an IPv4 stack IS configured > in its kernel. Patches needed to do that, if it doesn't already > work, would probably be accepted by the maintainers. Especially if > you provided a patch to the kernel networking test-cases, which > demonstrates the failure, and also demonstrates that your patch > eliminates the failure. Yeah, I am relying on so-far-unnamed co-conspirators to write those patches given my limited ability to write even remotely acceptable code. ;-) Still, i guess there is an argument to be had whether IPv4- compatible/IPv4 mapped addresses can be considered 'assigned' to an interface if the 'real' IPv4 is on there. With best regards, Tobias -- Dr.-Ing. Tobias Fiebig T +31 616 80 98 99 M tob...@fiebig.nl Pronouns: he/him/his _______________________________________________ Int-area mailing list -- int-area@ietf.org To unsubscribe send an email to int-area-le...@ietf.org
