serverguide: IP addresses "classes" definition is old/wrong

In the server guide one can read[0]:

"IP addresses are defined in classes whereby the bits making up an address correspond to network and host ranges. There are four principal address classes: A, B, C, and D. Class A IP addresses would typically be used in very large organizations and have the format network.host.host.host whereas the more common Class C address could be used in smaller organizations and is often used for individual host subscribers to Internet Service Providers. A Class C address would resemble network.network.network.host Class D addresses are special in that they are known as Multicast addresses. Some examples of valid IP addresses are 127.0.0.1 and 82.211.81.166."

That information is not only old, but wrong. If we considered Andrew S. Tanenbaum's "Computer Networks 4th edition", we would either fix that by saying that Class A networks are those ranged from 1.0.0.0 to 127.255.255.255, Class B from 128.0.0.0 to 191.255.255.255, etc, or just remove that paragraph.

To say that a "class A" network are those with a netmask of "network.host.host.host" is technically wrong, even though many years ago, before the invention of CIDR addresses, class A networks indeed had 8 bit netmasks. That was not what *defined* the class of a network, anyway.

My suggestion is to remove all that paragraph and not try to fix it, which could make it even more confusing. The classes allocation is not used anymore. In my brazilian portuguese edition or Dr. Tanenbaum's book you can find that in section 5.6.2, page 464. :-)

If the documentation is freezed, I'm sorry to have missed that, but it could be fixed for the next release, I guess.

http://doc.ubuntu.com/ubuntu/serverguide/C/tcpip.html