Re: WG action: 1 week to discuss LL18: Host

    Date:        Sat, 6 Sep 2003 07:12:01 -0700 (PDT)
    From:        Bernard Aboba <[email protected]>
    Message-ID:  <[email protected]>

  | I'd suggest that perhaps the multi-homing issue should be explicitly
  | addressed in the paragraph.  For example:

I think I agree with that, but I don't think I'd do it this way...

  | "If the host has no appropriate routable source address, but is not
  | multi-homed, then it SHOULD ARP for the destination address and then send
  | its packet, with a link-local source IP address and a routable destination
  | IP address, directly to the destination on the same physical link."

Rather, I think I'd say something like

	A multi-homed host needs to select an outgoing interface.
	Details of that process are beyond the scope of this specification.
	After selecting an interface, the multi-homed host should send
	packets involving link-local addresses as specified in this
	document, as if the selected interface were the host's only interface.

Otherwise, there's no mention at all of what multi-homed hosts should do
(not even that the doc isn't giving any advice) - they're simply being
excluded.

How a multi-homed host picks which interface to use is (for now) someone
else's problem - but having selected interface N, if it has just LL
addresses to use, then it should (SHOULD/MUST as determined) be doing the
same ARP for any random unknown destination as it would were it a singly
homed host.

Note that the above says absolutely nothing about how the interface should
be selected, nor what information the host can use to make that selection,
nor in which order anything is done - it doesn't even preclude sending an
ARP out every interface and picking one from which a reply is received.

kre