--On Monday, January 29, 2018 10:11 -0800 Ted Hardie <ted.i...@gmail.com> wrote:
> Sorry for the delay in replying; I was returning from the QUIC > meeting in Melbourne. > > On Thu, Jan 25, 2018 at 7:13 AM, John C Klensin > <john-i...@jck.com> wrote: >> --On Thursday, January 25, 2018 14:55 +0100 Frank Ellermann >> <hmdmhdfmhdjmzdtjmzdtzktdkzt...@gmail.com> wrote: >> >> > On 22 January 2018, Ted Hardie wrote: >> >> If you have thoughts >> > >> > Maybe "obsoletes 4698" instead of only "updates 4698" would >> > be clearer for readers of RFC 4698. >> >> Ted, >> >> Unless there are considerations that I don't understand, I >> agree with Frank and would go a step further. While the >> document indicates that IRIS was not actually deployed for >> address registry usage, as far as I know it has not been >> deployed for anything else either and has become part of the >> wreckage along the path to try to replace Whois for registry >> database use. >> >> If the intent here is to say "we have given up on IRIS" >> > > This came up in the context of the IAB trying to work out how > useful some of the existing delegations in .arpa are, and the > current draft reflects that. Updating RFC 4698 was, in other > words, the simplest thing we could do. Given the feedback, > I'm fine with updating that to "obsoletes RFC 4698", since the > data support the notion that this is not currently in use. > > I don't have the data to support a broader statement about > IRIS, as I know that there was some deployment at one time. I > would be fine seeing a deprecation if one is warranted, in > other words, but I'm not sure I can put one forward to the > community. > > Do you object to obsoleting just RFC 4698 by this document? Object, not really. I do see it as creating something of a silly state in which we leave the protocol apparently active and recommended while eliminating a key facility for utilizing it in a particular way should one decide to do so. The only thing that makes the address registries different in this regard is that the number of possible uses is small and easily identified. One of your comments above helps identify what concerns me about this document, so let me take a step back and address that. So the IAB decides that it should do a review of "how useful some of the existing delegations in .arpa are". Seems worthwhile to me although I might wonder whether the Internet and the various protocol relationships were in such good shape that the IAB should prioritize that work, But then I'd expect an I-D that said that the IAB had conducted that review and was, e.g., elimination all of the subdomains/ registries that were unused or that supported obsolete protocols (or particular applications of protocols) and obsoleted the standards-track documents creating those registries. Perfectly orderly. But, here, the document seems to single out one particular subtree and a single application of a protocol that has gotten little or no traction in it more general form. It that is the topic, I think the IAB is under some obligation to explore that protocol and its present-day utility and active implementations a bit further (especially in the light of Andy's note) or, if that is too burdensome, to include a (very brief) explanation about why IRIS itself is not being retired and/or why the IAB doesn't consider it useful to examine that question. Disclaimer: My position on this issue is conditioned by part of my view of the role of the IAB in current times. That view is that the IAB should be functioning as a sort of architectural think tank and, (at least at the conceptual level), an oversight body. That, in turn, implies that the IAB should be looking at whole-system issues, specifically including how particular protocols and sets of protocols interact with what has recently started being called the "Internet ecosystem". While it isn't my personal highest priority, I think the IAB's taking on a review of RFC 3172 and clean-out of registries that are no longer relevant would be a find idea. Conversely, if the IAB decides at another look --at a system-wide level-- at registry databases of many or all flavors is in order, I'd consider that a worthy task, even if it focused on cleaning up the options that have never taken off, whether you went back to Whois++ (which has been declared obsolete) or Rwhois (which hasn't). But to clean out a single application (and registry) or a single protocol just seems to me like a poor use of time and a likely source of confusion for those who might decide to read and think about the IRIS protocol documents. My vocabulary is somewhat different from that used by Brian and the IAB in 2008, but I believe it is consistent with RFC 2850. If the community no longer believes in that vision of the IAB, my view may be irrelevant, but a revision or replacement for RFC 2850 is probably in order. best, john _______________________________________________ regext mailing list regext@ietf.org https://www.ietf.org/mailman/listinfo/regext
