On Fri, 2003-08-22 at 23:40, Tony Hain wrote:
> > It would be nice to see some of this happen. While the bulk
> > of the work is a matter for another WG, the addressing issue
> > clearly isn't.
>
> To some degree the discussion is restricting the potential solution space by
> assuming that one has to choose between LL, PI, ??? .
I didn't mean to imply that this would be the only address assigned to
the node. What I wanted to ask was, what type of address should be used
in ad-hoc situations, regardless of what other addresses a node might
have.
> The state that was not
> discussed was arriving at 2 because the router crashed, while a new node
> comes up thinking the network must be in 1 because there is no RA. This is
> not a problem as long as existing nodes keep both the LL & RA derived
> prefixes, because the new node can talk to their LL address. The problem is
> that if the existing nodes are forced to choose a single address, they
> either break the existing app when the router crashes, or they refuse to
> talk to new nodes until the router comes back up or the lifetimes expire.
I agree. Forcing a node to choose a single address is unworkable, as it
would require groups of communicating nodes to execute synchronized
transitions between addressing states. In ad-hoc scenarios the only
stable perspective is that of a single node. The environment consists of
other nodes and is constantly changing.
MikaL
--------------------------------------------------------------------
IETF IPng Working Group Mailing List
IPng Home Page: http://playground.sun.com/ipng
FTP archive: ftp://playground.sun.com/pub/ipng
Direct all administrative requests to [EMAIL PROTECTED]
--------------------------------------------------------------------
