You remove a possible functionality (which is provided by other players) for the gain of couple of bytes in a static table and a slightly shorter man page.
Of course there private data field will become a bit shorter, too.

Why do you keep the definitions of these constants in ng_iface.h? Are there any other consumers?

Îf you want to stay in touch with the kernel capabilities, please consider a dynamic hook management which obtains the valid address family names from the real kernel table. This way, any change in the kernel supported protocols would be automatically available on the ng_iface.

In D25862#572900, @lutz_donnerhacke.de wrote:

What's the reason behind this proposal?

The documentation is incorrect. I wish to correct it. And if I do that I see no reason to keep around dead hooks either, so I removed them.

You remove a possible functionality (which is provided by other players) for the gain of couple of bytes in a static table and a slightly shorter man page.
Of course there private data field will become a bit shorter, too.

The actual functionality was removed a long time ago and has not come back. Sure, this is not a significant change, but I noticed the disparity when reading ng_iface and see no reason to leave it that way.

Why do you keep the definitions of these constants in ng_iface.h?

I missed them, thanks.

Are there any other consumers?

Nothing in the base system it seems.

Îf you want to stay in touch with the kernel capabilities, please consider a dynamic hook management which obtains the valid address family names from the real kernel table. This way, any change in the kernel supported protocols would be automatically available on the ng_iface.