Many home netgear routers have a SIP ALG bug which causes corruption and breaks SIP calls even when the port forwarding is correct.

At least it's not a recursive acronym.

NAT is useful even if address space wasn't a problem. For example, I routinely forward SSH and SMTP connections using simple iptables forwarding. (Side Note: it is extremely disappointing that Linux IPv6 stack hasn't even implemented this port forward functionality).

Even if NAT is out of the picture, you ignored my point about SIP being impossible to firewall without custom code (an ALG). Presumably firewalls will have greater importance once (in the distant future) NAT gateways are gone.

There is no technical reason that SIP requires more than one (static) port, it is a design decision that will make SIP forever difficult to use.

Take note how relatively well P2P apps work behind NAT considering the limitations.

Firstly, you still haven't addressed the firewall/ALG issue with SIP I brought up twice, which still exists without NAT.

Secondly, whether you choose to acknowledge it or not, port forwarding does have valid uses outside the scope of mitigating limited number IP addresses under IPv4.

One example is load distribution, the gateway router should have the ability to redirect inbound packets to multiple servers. This is an example of something NAT is good at.

Another example is the ability to use one domain name for multiple services. Domain names (typically) resolve to one IP address, without port forwarding/NAT, you're essentially forcing all daemons for one domain name to reside on a single server. This is a stupid limitation. For example, I may want 'domain.com' to have HTTP, FTP, SIP, all residing on separate servers. NAT/port forwarding on the gateway is far simpler, more efficient, and more scalable than forcing all protocols to be handled on one server.

Another example is a company moving servers to a new IP/location may want to port forward the requests on the old IP to the new servers' IP.

A similar example is when a server has maintenance downtime, the admin can forward requests at the gateway to an alternate server until the original server is ready. The port forwarding switch is instantaneous.

I definitely understand the motivation for IPv6 and eliminating the dependence on NAT. However, it seems you've overstated your case that it is never useful.

The need for firewalls is really a separate debate.

You could say "NAT has no use for HTTP" or "NAT has no use for SSH", since NAT has nothing to do with these protocols either. Somehow I allowed myself to veer off from talking about SIP to talking about NAT, but these are two separate topics. My original point was that protocols which use a single port are much more manageable.

Why not? Shouldn't that decision be left to the sysadmins? Who are we to set policy?

You've never used xlite or another voip/im client to dial "mydomain.com" or "###@mydomain.com"? Not only does this work, it's highly practical.

If ever it became popular, I'm sure most businesses would like to use their domain name for VOIP calls as well as web hosting. Why should we require that a domain name can only be used for services running on a single server? There is no reason to impose limits like this.

Yes of course, but you could port forward the old IP to eliminate the DNS prorogation downtime. It's one solution we have today you think we shouldn't have, but why not?

I understand your agenda against IPv4, but your beef with port forwarding is exagerated. Regarding the load balancers mentioned above? Please explain why these shouldn't be permitted on IPv6?

I already know ipv4 is a nightmare, but port forwarding is useful even in corporate environments where IPv4 was never a limitation....you are simply overstating your case.

Parent's point still stands. The GNU greatest hits came out years ago.

Exactly. It is like everybody using MS Office and your OpenOffice exchange is broken.
Everybody uses skype because it works, it passes through routers and works even inside big firewalled company networks.