On Feb 16, 2017, at 12:52 PM, Mark Smith <markzzzsmith(_at_)gmail(_dot_)com>
wrote:
So what is your objection to having a well-known addressing structure?
I think the big problem is that it's well-known except when it isn't. We don't
always have a 64 bit IID; we explicitly allow for prefixes like /127 (RFC 6164)
etc in certain cases. It's a CIDR prefix (RFC 7608), meaning that prefixes in
routing can be any length that makes sense operationally. 7608 was written in
large part to advise chip vendors that prefix lengths longer than /64 needed to
be OK.
So it isn't so very well-known. The 64 bit IID is a convention, not hard and
fast. I would differ from Randy's characterization because of the kernel truth
of RFC 7608, and with yours for the same reason.
And we seem to repeat this debate periodically. It gets boring.