Brian Haberman <[EMAIL PROTECTED]> writes: > Hi Ed,
> Ed Jankiewicz wrote: > > Speaking as an individual, but drawing on my experience as a contributor > > to one of the profile specifications you mention, it seems this draft > > could be one of the following: > > > > 1. a roadmap/reading list for IPv6 implementors and evaluators (thus > > informational, and without any RFC 2119 language) > Would this type of document be a single laundry list of IPv6-related > specs or would it try and narrow down the suggested reading list based > on some classification (e.g., router, host, etc.)? I think it should narrow down the list to the most important ones. Not only the "must" ones, but any that an implementor should be aware of and may want to implement. But it would be useful if this document was a general catalog of all the important IPv6 RFCs an implementor would want to be aware of. So, I don't think it needs to include *all* IPv6 RFCs, but there is little harm in listing them just for completeness. But I would like to see a fair amount of context/motivation for we recommend as a SHOULD or MUST. > > 2. an applicability statement as you suggest (informational or > > standards track) > The question I have with this approach is whether it will help stack > developers, especially those who develop a single IPv6 stack that gets > used in a variety of deployment scenarios (i.e., servers, clients, > caches, ...). It needs to help implementors. One problem with just having a "SHOULD implement" is that it doesn't include much context about what type of environments an RFC is useful in. Consider RFC 4294, section 7, on Mobile IP: > 7. Mobile IP > > The Mobile IPv6 [RFC-3775] specification defines requirements for the > following types of nodes: > > - mobile nodes > > - correspondent nodes with support for route optimization > > - home agents > > - all IPv6 routers > > Hosts MAY support mobile node functionality described in Section 8.5 > of [RFC-3775], including support of generic packet tunneling [RFC- > 2473] and secure home agent communications [RFC-3776]. > > Hosts SHOULD support route optimization requirements for > correspondent nodes described in Section 8.2 of [RFC-3775]. > > Routers SHOULD support the generic mobility-related requirements for > all IPv6 routers described in Section 8.3 of [RFC-3775]. Routers MAY > support the home agent functionality described in Section 8.4 of > [RFC-3775], including support of [RFC-2473] and [RFC-3776]. There is very little useful explanation above. And, IMO, the SHOULD recommendation to implement Route Optimization was (and probably still is) premature. But others looking just at the profile do not see anything to suggest that Route Optimization is something that isn't critical for general IPv6 nodes. > > 3. a normative definition of the IPv6 node requirements (standards > > track, inclusion by reference of current IPv6 standards with RFC 2119 > > language) > This type of document was discussed when the first Node Reqs draft was > written. I *believe* that the consensus at the time was that such an > approach was infeasible for any number of reasons. I'm not terribly thrilled with using 2119 language to show requirements. The market has already kind of decided (to a large extent) what it thinks is crucial to implement. Also, we would end up targetting a mythical "typical" node, of which only 80% of the devices really fit that model. The end result could easily be that some devices won't adhere to some MUSTs anyway, because they have special requirements. Think of what effect mandating (say) IKEv2 on all nodes would achieve. > I want to see a document that is the most useful to the community, > regardless of whether it takes more work or not. I agree! And FWIW, I'm willing to put some cycles into such a document, whether reviewing or providing text. My interest here comes largely from having watched the development of a couple of external IPv6 profiles, and I think the current node requirements document is much less useful than it could/should be. And, the node requirements is the way the IETF is stating what it thinks are the key IPv6 RFCs. Thomas -------------------------------------------------------------------- IETF IPv6 working group mailing list [email protected] Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
