-L<NetAddr::IP> objects have netmasks. When using them to represent
-range endpoints, be sure to set the netmask to I<zero> so that math on
-the address doesn't stop at the subnet boundary. (The default is /32,
-which doesn't work very well. Address ranges ignore subnet boundaries.)
+L<NetAddr::IP> objects have netmasks. They also have overloaded operators
+for addition and subtraction, but those have range limitations when comparing
+addresses. (An IPv4 address is effectively a uint32; the difference
+between two IPv4 addresses is the same range, but signed.) In later versions
+of the library the C<bigint> method can be used as a workaround, but
+otherwise it's not safe to subtract two addresses that might differ in the
+first bit of the first octet.