Begin forwarded message:
> The following errata report has been submitted for RFC6147,
> "DNS64: DNS Extensions for Network Address Translation from IPv6 Clients to
> IPv4 Servers".
> Original Text
> -------------
> An implementation SHOULD include the ::ffff/96 network in that range by
> default.
> Corrected Text
> --------------
> An implementation SHOULD include the ::ffff:0:0/96 network in that range by
> default.
We had some private discussions and there was a difference of opinion whether
leaving out the zero part of an IPv6 prefix is ok, same as in IPv4. For example:
10/8
2001:470:1f0b:1289/64
Doing this doesn't seem to be harmful even if systems may not accept these
prefixes. Except in the case from the DNS64 RFC, where leaving out zeros on
both sides in ::ffff/96 makes it impossible to tell which 16 bits are the one
bits. To add insult to injury, ::ffff/96 is in fact a somewhat valid prefix,
but not the one we had in mind, with unexpected results:
$ sudo ifconfig gif0 inet6 ::ffff/96
$ ifconfig gif0
gif0: flags=8011<UP,POINTOPOINT,MULTICAST> mtu 1280
inet6 ::0.0.255.255 prefixlen 96
And the existing guidance in http://tools.ietf.org/html/rfc5952#section-7
dismisses that prefixes have a problem.
Thoughts?
My position is that prefixes should always end in :: so the address part is a
valid address and the same code can be used to parse addresses, prefixes and
addresses with a prefix length.
--------------------------------------------------------------------
IETF IPv6 working group mailing list
[email protected]
Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------
