Jason,
we need to get rid of site-locals. merely renaming them as private use
addresses wouldn't solve any of their problems. there's no advantage to
moving to IPv6 if it repeats the RFC 1918 mistake.
Keith
On Sat, 26 Apr 2003 12:42:55 -0400
Jason Hunt <leth(_at_)primus(_dot_)ca> wrote:
I've done a bit more thinking regarding my previous post. I am going to
restate the things that I suggested, in case my previous message left
anything un-clear.
I would like to suggest that:
- The idea of scopes be removed from unicast addresses. Why add the
complexity of scopes? Is it not better to keep things simple?
- The link-local address space (FE80::/10), or some other space, be
available as "private use" address space, similar to what RFC 1918 is
to IPv4. I realize that the current definition of the local-use
addresses already provides this, but since I am suggesting to change
that definition it seemed necessary to make this point.
- All interfaces be required to have at least one unicast address
assigned to them, instead of being required to have a link-local
address in addition to any other addresses. If an interface is not
configured with an address, and the host is unable to obtain an
address from a DHCP server (or some other dynamic configuration
protocol) for that interface, then the interface will be auto-
configured with an address from the above-mentioned "private use"
space (FE80::/10 or otherwise).
Can anyone point out any practical scenarios for scoped addresses to
be required, which could not be dealt with by having a "private use"
address space available? After reading the discussion on site-local
addresses, I think that unicast address scopes may be un-necessary. If
I happen to be mis-understanding something, I welcome any explanations
or pointers to previous discussions.