Re: issue 5: zero address set

On Mon, 2004-11-29 at 09:04, Jari Arkko wrote:
> Francis Dupont wrote:
> > I don't believe zero address set support is so useful but I have
> > a little concern about it: when a node has really no available
> > addresses I prefer its peer knows this in place of just
> > keep the last usable address and suggest it does still work...
> > Of course, this situation should be temporary.
> 
> Right. I have a sense that we have difficulties to
> avoid temporary disconnects no matter what we do...
> often in wireless we don't get to choose whether we
> tell the other nodes that we are going to switch or
> go out of coverage; we get a problem and the protocols
> have to live with that.

I think there's a difference here between "mobile" (phone) and "nomadic"
(laptop) operation; in many cases, the nomad can do a graceful shutdown
prior to becoming disconnected.

That graceful shutdown means that the peer can be a better network
citizen (avoiding sending useless packets which may bother the next
tenant of the nomad's former address).

Here's an approach which makes zero address support effectively
optional:

Implementations which do support it must be prepared to cope with the
peer losing state while they were disconnected.

Implementations which don't want to support it treat it as if it were a
disconnect operation.

A humorous way to look at it: it's the "Terminator option" ("I'll be
back") to the disconnect operation...

							- Bill