Feature deaggregate(IP)

[martin-schmidt]

Takes IP and returns Array of Subnets.

   IP.new('1.2.0.0').deaggregate(IP.new('1.3.255.255'))
     => [#<IP::V4 1.2.0.0/15>]
   IP.new('1.2.0.0').deaggregate(IP.new('1.4.255.255'))
     => [#<IP::V4 1.2.0.0/15>, #<IP::V4 1.4.0.0/16>]
   IP.new('2001:db8:85a3:8d3::').deaggregate(IP.new('2001:0db8:85a3:08d3:ffff:ffff:ffff:ffff'))
     => [#<IP::V6 2001:db8:85a3:8d3::/64>]
   IP.new('2001:db8:85a3:8d3::').deaggregate(IP.new('2001:db8:85a3:8d3:1::'))
     => [#<IP::V6 2001:db8:85a3:8d3::/80>, #<IP::V6 2001:db8:85a3:8d3:1::>]

[candlerb]

I had to think a bit about what this is trying to achieve.

What I believe it is doing is taking an inclusive range of IP addresses (ip1..ip2) and returning the smallest number of prefix-aligned subnets which covers that range. So I would argue what it is actually doing is aggregating, not de-aggregating.

There is an existing to_range function which is almost the inverse of this.

So how about a class method on IP which takes a range, and returns an array of IP objects which covers that range? I'm not sure exactly what that should be called though.

[martin-schmidt]

What I believe it is doing is taking an inclusive range of IP addresses (ip1..ip2) and returning the smallest number of prefix-aligned subnets which covers that range. So I would argue what it is actually doing is aggregating, not de-aggregating.

There is an existing to_range function which is almost the inverse of this.

That's exactly what it's doing. It's called deaggregate in ipcalc, so I stuck with this to be consistent and because I haven't had a better idea.

It was the easiest way for me not to use ranges or a class method, because I use it with already initialized /32 objects, but ranges may be the proper way to do it in some other use cases.