Remi, >> because no-one will ever do this? > > Assuming that details that follow mean that an expert can configure a node > with an address that isn't unauthorized by any RFC, and in particular a > 4rd-reserved address, that's acknowledged. > But nothing specific needs to be done, in 4rd not more than in any > specification, to make hand-configured unauthorized addresses to work > properly. > > The difference between 4rd and MAP, in this respect, is as you know that MAP > addresses can conflict with *authorized* host addresses. This is AFAIK the > reason for your suggesting that sites that use subnet 0 be renumbered to > support MAP.
just like any interface address can conflict with any other. MAP uses a single address out of the first subnet in a delegation. the probability of collision even if it was shared with native hosts is reasonably small. your view of the effect of reserving some bits in the middle of the interface-id for the purpose of 4rd seems quite overstated to me. nor do I follow your use of "authorized / unauthorized" interface-ids. the solution proposed in MAP may not be perfect. it is a tradeoff. 4rd suggests another tradeoff that in my view, has much wider consequences. take it to 6man. cheers, Ole _______________________________________________ Softwires mailing list [email protected] https://www.ietf.org/mailman/listinfo/softwires
