> In my opinion it is a bug and deserves a PR.
> 
> Someone (mycroft?) said something about it not being a bug because it's
> documented somewhere.  I don't see documentation for it anywhere;
> bind(2) points at the section 4 manpages, and neither ip(4) nor tcp(4)
> said anything about the structure of a sockaddr_in as far as I could
> tell, never mind what the semantics of the sin_zero array are.  tcp(4)
> does refer to inet(4), which lists the members of the structure, but
> does not give any semantics for any of them.

Actually inet(4) itself is a bug and deserves a PR.  sin_len was added many,
many years ago.

In fact one could argue that the `right' thing to do would be to remove the
sin_zero field altogether.  The reason the semantics for sin_zero are neither
documented nor obvious is that it has none; its presence is a historical
artifact, with the reason for its existence long gone.

I wonder what it would break to just remove it?

Dennis Ferguson