> Hmm. I think it is less desirable to break compatibility with the
arp(8) -f
> input file, but not tragic if we do so. Still, I'd like to avoid doing so
> without a clear need, and I haven't seen one.

maybe the syntax I proposed is not the best, but it would be nice to have
arp(8) use the existing ether-address to hostname mapping mechanism instead
of having to specify ether-addresses in multiple places.

    christian