Perhaps it is time to recall RFC 6982. While that RFC says nothing about setting hurdles (so that debate needs to continue) it does provide a lightweight way of tracking implementations and noting what bits of I-Ds have been implemented.
Yaron and I can report mixed results from the RFC so far: sometime there has been a degree of enthusiasm to report implementations (free marketing?) while other times implementations that we know are out there have sullenly refused to present details for inclusion in I-Ds. Draw your own conclusions :-) Adrian > -----Original Message----- > From: BESS [mailto:bess-boun...@ietf.org] On Behalf Of Thomas Morin > Sent: 26 November 2015 13:50 > To: bess@ietf.org > Subject: Re: [bess] Introducing a one-implementation requirement before WG > last calls > > Kireeti, Martin, > > Any data related to a serious implementation (even a beta without a > shipping date), would I think be convincing enough. > "release x.y @shipping date d" was simply provided as a possible answer, > not of the minimum requirement to be convincing. > > As Martin said we need more than "I am aware of an implementation". > I would say we also need more than "we plan to implement". > > There are cases where we may need more detailed data. > Take for instance the case of a draft that needs specific dataplane > support: would we accept as convincing enough an implementation on a > software-only platform, in the absence of any vendor planning to > implement on hardware ? > > -Thomas > > > > Martin Vigoureux : > > Hello Kireeti, > > > > thanks for your inputs. > > I understand the challenge that "release x.y @shipping date d" might > > pose. What we want, is to go beyond the "I am aware of an > > implementation" type of response. It might currently be sufficient > > with regards to the shepherd write-up question, but won't be any more > > if we introduce the requirement. We'd like to have tangible > > information. Giving "details on how much of the spec was implemented" > > is clearly going in that direction. > > > > -m > > > > Le 26/11/2015 02:26, Kireeti Kompella a écrit : > >> Sounds like a good idea to me. One tweak: having an official "release > >> x.y @shipping date d" is unlikely for a draft. The value of one > >> implementation (vs more) is that it shows that a spec is > >> implementable and reasonably complete. So, this should be the focus, > >> with details on how much of the spec was implemented. Shipping plans > >> should be totally optional. > >> > >> Note that even an experimental implementation takes effort, is likely > >> to become official, and shows a degree of seriousness of the part of > >> the implementor. Asking for greater commitment at WGLC is (imho) > >> asking too much. > >> > >> Kireeti > >> > >>> On Nov 24, 2015, at 01:03, Thomas Morin <thomas.mo...@orange.com> > >>> wrote: > >>> > >>> Hello everyone, > >>> > >>> Following the positive feedback received during BESS meeting in > >>> Yokohama about introducing a one-implementation requirement in BESS, > >>> we propose to do the following for future WG last calls: > >>> > >>> As a prerequisite before doing a working group last call on a > >>> document, the chairs will ask the working group for known > >>> implementations of the specifications; a relatively detailed level > >>> of information will be required (e.g. "release x.y of solution z > >>> shipping date d", "all features implemented", "partial > >>> implementation only", etc.) and everyone will be invited to reply > >>> (not only co-authors of the specifications); the chairs will then do > >>> the working group last call if satisfying information was provided > >>> on at least one implementation. > >>> > >>> We are open for comments on this proposal until December 7th. > >>> > >>> Martin & Thomas > > _______________________________________________ > BESS mailing list > BESS@ietf.org > https://www.ietf.org/mailman/listinfo/bess _______________________________________________ BESS mailing list BESS@ietf.org https://www.ietf.org/mailman/listinfo/bess
