multiple prefixes

Lorenzo Colitti lorenzo at
Wed Feb 13 02:05:52 CET 2013

On Wed, Feb 13, 2013 at 3:27 AM, Doug Barton <dougb at> wrote:

>> An example I have direct experience of is the code that does connection
>> establishment libjingle, used by Google video chat. Being able to rely
>> on the fact that the IPv6 address assigned to your interface is, in
>> fact, your IPv6 address greatly simplifies the code. By comparison,
>> firewall traversal is trivial.
> Fair enough, but I live in a world where that issue has already been
> addressed, with multiple different solutions available.

Even if there is code available, you're still talking about development and
testing cost (and decreased reliability in corner cases), in all non
client-server applications, forever. It doesn't have to be that way.

Would life be easier if such solutions were not necessary? Sure! Are we
> ever again going to live in a world where they are not? Nope.

I disagree. Here's an alternative scenario: enough of the userbase of these
applications has non-NPTed IPv6. There is insufficient incentive for the
developers to support NPT. Thus, NPT is not supported by the majority of

Since you use today's NATed IPv4 world as an example, here's an example
from that world of apps that don't bend over backwards to work everywhere:
AFAIK Skype (arguably the app with the best workaround) doesn't work behind
networks that don't allow TCP or UDP connections to the outside world but
only use HTTP proxies. Technically, it could probably be made to work
behind an HTTP proxy by implementing tunneling over HTTP, but it doesn't.
-------------- next part --------------
An HTML attachment was scrubbed...

More information about the ipv6-ops mailing list