On 2011-01-06 02:15, RJ Atkinson wrote: ... > A) Prohibiting new IPv6 Extension Headers outright, > as Joel has repeatedly suggested. This removes the > narrow case where RFC-2460 allows Extension Headers, > so the I-D would be an Update to RFC-2460.
My reaction is that this is going too far, and isn't actually necessary. We have words in 2460 that already make it quite hard to define a new header. We have running code proof (in the form of non-existent IPv4 options) that it's in practice impractical to deploy such things, for exactly the reasons we've been discussing (routers are not forward-compatible, and firewalls block unknowns by default, both being partially embedded in silicon). But who knows what the future may bring, in 20 or 40 years from now? So I am more inclined to a SHOULD NOT approach; I think I'm agreeing with a somewhat stronger version of Suresh's proposed paragraph. At least, add ...accompanied by ... and a convincing explanation of a successful progressive deployment model for the proposed new header. Brian -------------------------------------------------------------------- IETF IPv6 working group mailing list ipv6@ietf.org Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
