I wrote:
> There is no subset type that corresponds to "valid host addresses
> only" --- if there were, it would be a subset of INET but would have
> no valid values in common with CIDR.
I take that back --- CIDR accepts w.x.y.z/32 for any w.x.y.z, which
would include valid host addresses. (But perhaps it should only
accept netmasks shorter than 32 bits? Not sure if "CIDR" is commonly
understood to be network specs only, or network and host specs.)
regards, tom lane