Re: [homenet] Configuration must not be carried by the routing protocol

[Juliusz Chroboczek <[e-mail]>]

>> However a profusion of protocols exist, and having in a configuration
>> protocol the ability to announce and/or select a routing protocol
>> would be the most flexible.

> I don't want this.
> Then we get into MUST implement routing protocols, and SHOULD implement
> routing protocols, etc...

We define The One Homenet routing protocol.  Full stop.

We don't want any options in Homenet (RFC 1869, page 2), we want
a clean protocol stack that guarantees interoperability.  However,
this should not prevent us from designing it as a modular collection
of protocols (using already-deployed protocols whenever possible)
rather than throwing everything including the kitchen sink into one
monstruous protocol that nobody is ever able to reimplement from scratch.

> I am claiming that the decision, "can we trust this configuration
> information" may be linked to "what route this this configuration
> information arrive".

I don't buy this.  Actually, I don't buy any claims about security
that don't include a clearly defined threat model and don't show how
the proposed policy applies to the threat model.

-- Juliusz