chaotic serial numbers

[Simon Waters]

Michael Kjorling wrote:
> 
> I bet it has to do with the fact that the sign bit (quite
> conviniently, bit 31 as in 31..0) is supposed to be ignored, but I
> don't know the exact mechanisms behind it.

RFC1982 remembers so you don't have to remember it, although
this part in section 3.1 that says (I paraphrase) " you can
increment a serial by a positive integer n where n is from
the range [0 .. (2^(SERIAL_BITS - 1) - 1)]" was never proof
read by a mathematician I suspect *8-)

Basically a positive 32 bit number, that wraps around,
increments restricted as above (Is it an increment if you
add zero?!). Oh and don't make the serial number zero.

I vaguely recall that the real world can be more complex
than the standard, but that isn't a surprise (or at least
shouldn't be).