[dns-wg] Reverse DNS operational considerations in an IPv6 environment

I guess this is a question suited for this WG. 
References to other technical forums that can address this, or any pointers to 
established or not operational guidelines, are highly welcome.

The question has to do with the provisioning of the reverse DNS entries 
of customers of a broaband provider in an IPv6 environment (PTR records in 
ip6.arpa). In the IPv4 world, each user receives mainly a single IP. 
There are dynamic and static users and most providers (including us) have 
pre-populated PTR records in their allocations in in-addr.arpa. PTR records 
can point to different namespaces depending on the address space use (eg 
static.<domainname> and dynamic.<domainname> for static and dynamic customers 
respectively).

In the IPv6 case the problem is obvious because of the huge IPv6 address 
space. Since each customer can receive a /56, this is an assignment of 2^72 
possible /128 addresses which is a tremendously HUGE number. We cannot 
possibly have pre-populated any zone with that many entries and this is just 
a single customer.
The main question therefore is: how can we address this issue, if we need to 
have reverse DNS resolution for IPv6 ranges? The first thing that comes to my
mind is using wildcard records in BIND, or skipping reverse DNS alltogether.
Do you know of any operational recommendations?

Thanks,

Kostas Zorbadelos