Dual stack IPv6 for IPv4 depletion

Fri Jul 10 02:08:32 UTC 2015

> On Jul 9, 2015, at 16:28 , Ricky Beam <jfbeam at gmail.com> wrote:
> 
> On Thu, 09 Jul 2015 19:08:56 -0400, Owen DeLong <owen at delong.com> wrote:
>> the reality I’m trying to point out is that application developers make assumptions based
>> on the commonly deployed environment that they expect in the world.
> 
> Partially. It's also a matter of the software guys not having any clue what-so-ever w.r.t. networking. In this case, APPLE designed Bonjour to not cross network boundaries. Idiotic, but it allows them to sell "servers" that do the cross-network routing.

Actually it’s not a design problem in IPv6. A simple tweak to the software to send to ff05::<group> instead of ff02::<group> or better yet, allow the user to edit the scope in System Preferences is all that is really needed.

However, in IPv4, mDNS/Bonjour (and mDNS is uPNP’s fault, not Apple’s to the best of my knowledge) use broadcast packets and that’s a design flaw. However, hard to argue with the choice since multicast, especially cross-router multicast is pretty much busted in any sort of home gateway in IPv4 anyway.

>> If we create a limited environment, then that is what they will code to.
> 
> They will code to what they understand, what "works for them", and what their users report "works for me". We will always end up with "substandard" quality because they have little (or no) understanding of how networking does it's thing.
> 
> (And then marketing, and legal will step in and pooh on it even more.)

OK… Clearly you are determined to let cynicism and avoidance drive your ideas here. I can’t help that.

Hopefully enough others will try to make the internet more useful as we move forward and hand out larger end-site prefixes.

Owen