Re: 6to4 using ::FFFF:0000:0000/96 (mail.comcast.net AAAA record weirdness)

[Brian E Carpenter <brian.e.carpenter@gmail.com>]

On 2008-01-29 01:53, Iljitsch van Beijnum wrote:
> On 28 jan 2008, at 11:17, Rémi Després wrote:
> 
>>>> << Applications may use PF_INET6 sockets to open TCP connections to
>>>> IPv4
>>>>   nodes, or send UDP packets to IPv4 nodes, by simply encoding the
>>>>   destination's IPv4 address as an IPv4-mapped IPv6 address, and
>>>>   passing that address, within a sockaddr_in6 structure, in the
>>>>   connect() or sendto() call. >>
>>> How important is this feature in practice?  Could we deprecate it?
>> IMO deprecating it would be a bad step backward!
> 
> 
> Indeed.
> 
> I can't say that I spend a lot of time writing applications, but I
> consider having a single interface to talk IPv4 and IPv6 extremely
> important. Without that, it's very easy to have bugs that only come up
> in one IP version and it is of course extra work to have two sets of
> socket calls. Worse, it's not possible to create IP version agnostic
> applications.

I agree. That's why I've just come to realise the disastrous effect of
overloading IPv4-mapped addresses for both RFC 2553 and NAT-PT/SIIT usage.
We blew it.

Unfortunately, the above "may" isn't even a normative MAY, let alone
a SHOULD or MUST.

   Brian