On Aug 10, 2010, at 14:57, Michael Richardson wrote: > The rational for lots of bits would be to > encourage random allocation such that in the event that two > uncoordinated ROLL networks happened to merge (even for a few > minutes!!!) that the likelyhood of cross talk would be reduced.
Hmm. If this is intended as a routine way to handle the situation, the likelihood is still way to high. (If this is just intended as a mitigation to increase robustness in a case that already is an error situation that is being actively avoided by other means, your argument might be more interesting.) > A merge like I describe here is not directly envisaged by the current > set of requirements in my opinion, but I think that requirement will > emerge in the future. DANGER. Adding complexity to a protocol to cater for requirements that don't exist yet leads to convoluted solutions. Never do that. (Note that this situation is different from the selection of one out of a set of same-complexity alternatives, where it is quite useful to conjecture about potential future requirements to generate a tie breaker. Note also that extensibility by itself is a known current requirement for any protocol, which is quite different from designing in all conceivable future extensions.) Gruesse, Carsten -------------------------------------------------------------------- IETF IPv6 working group mailing list ipv6@ietf.org Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------