Re: Application protocols and Address Translation

Eliot Lear wrote:
> 
> Patrik,
> 
> You are precisely correct, that the use of RFC-1918 address space was
> meant for those devices who were envisioned to never attempt
> communications either through the Internet or even beyond a single
> administrative realm.
> 
> That having been said, there seems one legitimate argument left for
> site-locals:
> 
> Totally disconnected networks.  

Insert "or intermittently connected with a variable prefix" and that's it.
But this argument has been beaten to death on the IPv6 lists.
The question is, what are the applications-specific arguments 
against translated addresses? What are the brokenness conditions
caused by translated addresses (and the associated statefulness)?

   Brian