Re: PS to IS question from plenary
Just to make sure we have good data, can we go back a few more years? Specifically, did we not previously have a restriction forbidding references FS-DS, and {FS,DS}-PS? RFC3967 was in Dec. 2004, but I thought that we had some other work more recently (2008?) that attempted to unjam things. What I'm wondering is if the 20 actions prior to 6410 was an anomaly, and really the historical rate of upgrade is really lower. A quick couple of seconds with mail search and grep says that in the four years preceding publication of RFC 6410, we made 28 actions. As noted earlier, 20 of those happened in the 22 months preceding the publication of RFC 6410. So in the 2+ years before that only 8 were produced. I'd love to have a graph that plots DS/FS/IS progression per year, over time. But I don't have time to make that today. Unless someone else wants to... Jari
Re: PS to IS question from plenary
Hi Dave, At 14:11 30-07-2013, Dave Crocker wrote: And, of course, if it can be reshashed, in the IETF it will be. Agreed. However the specification for the new two-stage model provides criteria for the second stage, and it does not include re-evaluating the technology or its details. Instead it focuses on adoption and use. RFC 5011 is one of the rare RFCs which does not have any erratum. The move wasn't problematic. It is not clear whether RFC 6891 is an odd case or symptomatic of the difficulties in updating a specification. The easier way not to have a re-evaluation is not to reopen the RFC (e.g. RFC 5011). Anything else means having two consensus calls on the text. draft-bradner-restore-proposed-00 proposes an alternative where the specification is not expected to be complete and comprehensive. Here's a minimum standards profile published in 2009: SMTP IETF STD 10 (RFC 821, 1869, 1870) DNS IETF STD 13 (RFC 1034, 1035) HTTP v1.1 (RFC 2616), URL (RFC 1738), URI (RFC 2396) IPv4 (RFC 791, 792, 919, 922, 1112) BGP4 (RFC 1771) The ex-IAB Chair explains that RFC 1035 is an Internet Standard. He then explains that an Experimental Standard (RFC 1183) updates that Internet Standard. The Internet Standard is updated by a Best Current Practices (RFC 6895). There is also a Proposed Standard ( RFC 1982) which updates the Internet Standard. He recommends RFC 2026 for further information. He receives an enquiry afterwards about immature specifications. He decides to ask the question at the plenary. The community thanks him for his valiant efforts. The alternative is to settle for Informational Standards as that requires less effort. Regards, -sm
Re: PS to IS question from plenary
--On Tuesday, July 30, 2013 16:41 +0200 IETF Chair ch...@ietf.org wrote: Last night there was a question in the plenary about how many PS-IS transitions have occurred since RFC 6410 was published in October 2011. That RFC changed the three-step standards process to two steps. There was also a question of how this compared to previous times before that RFC got approved. Looking at the timeframe from October 2011 to today (22 months), there have been four such protocol actions. These results are given by searching the IETF Announce mail archive: ... Prior to the publication of RFC 6410, in the preceding 22 months there were these 20 actions raising standards to either Draft Standard or Full Standard: ... I should insert here the Standard disclaimer about possibly faulty search methodology or records, misunderstanding the question, or the hasty interpretation of results. In particular, the above search was not easy on ARO, involved manual steps, and I might have easily missed something. And I wish I had been able to do a database query instead. Feel free to repeat verify my results... Jari, Thanks for this. Disclaimers and possible small classification errors aside and being careful to avoid making causal assumptions, I believe that the implication of the above is that there is no evidence that the 3 - 2 transition has increased the number of documents being moved or promoted out of Proposed Standard. If one were to assume a causal relationship and an absence of external confounding variates or processes, one might even conclude the the 3 - 2 transition has made things quite a lot worse. Conversely, it seems to me that one could argue that the change has made things better only by demonstrating the existence of a process that would have led to considerably fewer than four documents being moved out of Proposed Standard in the last 22 months in the absence of the change. While the apparently-significant reduction in documents moved out of Proposed Standard is far worse than we expected, is it time for Scott Bradner and myself to review draft-bradner-restore-proposed-00, issue a new version, and start a serious discussion about that model of a solution? Would be willing to sponsor such a draft or, if you prefer, organize a WG or equivalent to consider it? thanks, john
Re: PS to IS question from plenary
On 31/07/2013 06:27, John C Klensin wrote: --On Tuesday, July 30, 2013 16:41 +0200 IETF Chair ch...@ietf.org wrote: Last night there was a question in the plenary about how many PS-IS transitions have occurred since RFC 6410 was published in October 2011. That RFC changed the three-step standards process to two steps. There was also a question of how this compared to previous times before that RFC got approved. Looking at the timeframe from October 2011 to today (22 months), there have been four such protocol actions. These results are given by searching the IETF Announce mail archive: ... Prior to the publication of RFC 6410, in the preceding 22 months there were these 20 actions raising standards to either Draft Standard or Full Standard: ... I should insert here the Standard disclaimer about possibly faulty search methodology or records, misunderstanding the question, or the hasty interpretation of results. In particular, the above search was not easy on ARO, involved manual steps, and I might have easily missed something. And I wish I had been able to do a database query instead. Feel free to repeat verify my results... Jari, Thanks for this. Disclaimers and possible small classification errors aside and being careful to avoid making causal assumptions, I believe that the implication of the above is that there is no evidence that the 3 - 2 transition has increased the number of documents being moved or promoted out of Proposed Standard. If one were to assume a causal relationship and an absence of external confounding variates or processes, one might even conclude the the 3 - 2 transition has made things quite a lot worse. Conversely, it seems to me that one could argue that the change has made things better only by demonstrating the existence of a process that would have led to considerably fewer than four documents being moved out of Proposed Standard in the last 22 months in the absence of the change. While the apparently-significant reduction in documents moved out of Proposed Standard is far worse than we expected, is it time for Scott Bradner and myself to review draft-bradner-restore-proposed-00, issue a new version, and start a serious discussion about that model of a solution? Would be willing to sponsor such a draft or, if you prefer, organize a WG or equivalent to consider it? I would rephrase it as performing the inverse operation to RFC 4450. I don't think it needs to be a fully-fledged experiment under RFC 3933; just a large batch of PS-IS promotions in one go under RFC 6410. It is a good idea. Brian
Re: PS to IS question from plenary
At 11:27 30-07-2013, John C Klensin wrote: Disclaimers and possible small classification errors aside and being careful to avoid making causal assumptions, I believe that the implication of the above is that there is no evidence that the 3 - 2 transition has increased the number of documents being moved or promoted out of Proposed Standard. If one were to assume a causal relationship and an absence of external confounding variates or processes, one might even conclude the the 3 - 2 transition has made things quite a lot worse. Conversely, it seems to me that one could argue that the change has made things better only by demonstrating the existence of a process that would have led to considerably fewer than four documents being moved out of Proposed Standard in the last 22 months in the absence of the change. Changing the Internet Standards Process from three maturity levels to two is intended to create an environment where lessons from implementation and deployment experience are used to improve specifications. The change could be rated as a non-change if there were only four specification moved to Internet Standard since then. The hurdle in moving a specification (not a RFC) from PS to IS is that the draft goes through IESG Evaluation again. As for public review, it can be a hurdle too as the pervious discussions can be rehashed. A PS specification which sticks to what goes over the wire turns these hurdles into a lesser effort. draft-bradner-restore-proposed-00 proposes a nice fix and it might even help lessen time to publication. Regards, -sm P.S. Olaf asked the question to the correct body.
Re: PS to IS question from plenary
On 7/30/2013 11:00 PM, SM wrote: The hurdle in moving a specification (not a RFC) from PS to IS is that the draft goes through IESG Evaluation again. As for public review, it can be a hurdle too as the pervious discussions can be rehashed. And, of course, if it can be reshashed, in the IETF it will be. However the specification for the new two-stage model provides criteria for the second stage, and it does not include re-evaluating the technology or its details. Instead it focuses on adoption and use. No doubt, there will continue to be IESG and other IETF participants who insist on treating consideration of Full standard as an excuse to discuss first principles, but it's not within scope. Over time, as more specifications are processed to Full, the community will learn to be a bit more efficient about it. d/ -- Dave Crocker Brandenburg InternetWorking bbiw.net
Re: PS to IS question from plenary
Just to make sure we have good data, can we go back a few more years? Specifically, did we not previously have a restriction forbidding references FS-DS, and {FS,DS}-PS? RFC3967 was in Dec. 2004, but I thought that we had some other work more recently (2008?) that attempted to unjam things. What I'm wondering is if the 20 actions prior to 6410 was an anomaly, and really the historical rate of upgrade is really lower. I think it is also important to understand how many new PS could have been even ready at a given point for an upgrade. I don't have a good/simple metric for determining this, I think it takes some significant semi-expert review. Perhaps someone has an Information Science (?!) or Economics graduate student who wants to mine the data and write a paper :-) -- Michael Richardson mcr+i...@sandelman.ca, Sandelman Software Works pgp6bgZwX7Izy.pgp Description: PGP signature