IP Address tables are confusing

[ulf.bartelt@…]

Class Networks

A 10.0.0.0 B 172.16.0.0 through 172.31.0.255 C 192.168.0.0 through 192.168.255.255

What shall the table show?

Adresses in networks or network addresses?

Network adresses would be

Class Networks

A 10.0.0.0 B 172.16.0.0 through 172.31.0.0 C 192.168.0.0 through 192.168.255.0

[gerard@…]

Read the preceding sentence again:

make sure that the IP address is in the private network IP address range. Valid ranges are:

<table here>

The table lists all the valid IP addresses. Granted it lists the network and broadcast addresses on the outer fringes too. Maybe that ought to be changed.

[gerard@…]

Ignore my last comment. We don't show a valid range for the 10.0.0.0/8 network either.

I'm moving this to lfs-dev. Some more advanced network layout discussion is required.

[bryan@linuxfromscratch.org]

In a few minutes I'll attach a proposed patch to the current SVN book, which I think describes the valid networks better, and also explains how there are actually 256 individual /24 networks in the 192.168 range (and likewise for the 172.16-31 range). I wish I'd have seen this information myself, before trying to set a /16 subnet on a 192.168 network. Not that it's impossible, but it's not recommended.

I've included a link to Wikipedia's CIDR page, which I believe has a good explanation of exactly what the /XX notation means. Wikipedia probably isn't a great place to go for reference information in general, but this particular article looks correct. Of course, we can always just recommend that the user Google for "CIDR notation" if we want.

[bryan@linuxfromscratch.org]

Patch to update the IP address table, and make its explanation make more sense

[bdubbs@…]

Although it is not clear, I think the ticket is referring to Section 7.11, Customizing the /etc/hosts File.

In this section, FQDN is not defined.

The discussion in the book (and the patch) needs to avoid the idea of Classful routing completely. They are long depreciated. There is not a single system on the internet today that is not CIDR.

Instead of class, the most appropriate reference would be to the "normal" prefix such as:

Private Network Address Range   Normal Prefix  
10.0.0.1  - 10.255.255.254        8
172.x.0.1 - 172.x.255.254        16
192.y.0.1 - 192.168.y.254        24

x is any number in the range 16-31
y is any nymber in the range 0-255

Also, in the script, the following:

[192.168.1.1] [<HOSTNAME>.example.org] [HOSTNAME]

Would probably be more accurate as

<192.168.1.1> <HOSTNAME.example.org> [alias1] [alias2] ...

[bdubbs@…]

Oops, of course the line:

192.y.0.1 - 192.168.y.254 24

should be:

192.168.y.1 - 192.168.y.254 24

[bryan@linuxfromscratch.org]

Bruce, that sounds good to me; I agree that classful routing is deprecated pretty much everywhere. (And yes, I also think the issue is with section 7.11.)

I have a patch doing what you suggested (expanding the FQDN acronym the first time around, modifying the table, and adding a paragraph of explanation), coming up in a minute or two. This is an alternative to the patch I posted earlier.

I'm not quite sure what you meant by the change from square to angle brackets though (unless you meant to merely change HOSTNAME to alias1/alias2?), so I didn't do anything with that.

[bryan@linuxfromscratch.org]

Patch to current SVN with Bruce's proposed table and wording

[bdubbs@…]

The issue of the angle/square brackets is that the values in the angle brackets are required, but should be changed. The values in the square brackets are completely optional.

I realize that this is different from the Typography section in the preface. Perhaps its worth a more general discussion than in this ticket.

[Jeremy Huntwork]

Done as of r7501