[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: next (hopefully final) revision (08) of ERS - discussion about order of hashtrees by time and not binary
thank you for the explanation.
And I am very happy that you can live with the compromise we discussed.
Please accept my apologies in case I might have understood s.th. wrong.
And also thank you for making me aware of the mistake with the intended status of the draft. You are right it must of course be "Standard" (not "Experimental"); we will change the field immediately with the next version as I already mentioned in the review results.
Btw. independent of the status, as long as an I-D has not finally been released by the IETF it is always still work in progress as defined by RFC2026. So feedback and requests for change of an I-D are always accepted for discussion and welcome.
> -----Original Message-----
> From: Tilo Kienitz [mailto:tk-tlslist@xxxxxxxxxxxxxx]
> Sent: Monday, October 23, 2006 12:40 PM
> To: Tobias Gondrom
> Cc: ietf-ltans@xxxxxxx; André Damm-Goossens
> Subject: Re: next (hopefully final) revision (08) of ERS - discussion
> about order of hashtrees by time and not binary
> Hello Tobias,
> please let me explain the motivation of my bosses e-mail. The way
> you understood it, was not exactly what he meant.
> Our first implementation matched the ERS draft. But then we noticed
> that binary sorting throws away some information that could be
> useful: The order in which the documents came into the archive.
> Therefore we changed our implementation to respect the order of the
> archiving. At the same time I started the discussion on this list.
> I can assure you that we are not afraid of the work to change our
> implementation again. We do not want to change it, because we believe
> that we chose the appropriate approach to meet our customer's
> requirements. ERS is no standard yet, right now it's intended by you
> to get "status: Experimental". So it will still be in a state where
> implementators may find that a 95% perfect draft could be improved
> by adding an OPTIONAL feature.
> Kind regards
> Tobias Gondrom wrote:
> > Hello all,
> > after a short and very good email exchange with Tilo yesterday evening,
> > we agreed to the status that we should address the request and
> > discussion for an option to order hashtrees by time of archiving
> > (instead of binary order) as an additional option for a future revision
> > of ERS (version 2) and will proceed with the current ERS as planned to
> > come to a release.
> > Thanks, Tobias
> > Ps.: please note that I also received an email from Mr. Damm-Goossen
> > (manager of the company of Tilo) who also asked to change the
> > specification so that it matches their current implementation. His main
> > reason is because they've already rolled out their own proprietary
> > solution to some customers and they would have to change this if we
> > would not adjust our WG draft to match their implementation. At the
> > moment this argument is only motivating me more to finally get ERS
> > released as standard ASAP, and not to change the draft.
> > 1. arguments on the mailing list are conclusive: Robert's arguments and
> > the other discussions give clear reasons why the draft (with binary
> > order of hash trees) is the right and best approach. (binary order is a
> > reliable criteria, the time of archiving is NOT, it is s.th. that can
> > neither be guaranteed nor really relied on - detailed discussion on
> > mailinglist at May 16th and 17th this year.)
> > 2. The request to change a draft to match a proprietary implementation
> > because there are deployments by one individual company who didn't care
> > about the draft at the design time and now wants to reduce any work
> > implied by their individual approach can not be a valid argument to
> > change a draft. If the WG/IETF would follow such a path we would never
> > be able to agree on a standard at all.
> >>-----Original Message-----
> >>From: owner-ietf-ltans@xxxxxxxxxxxx
> > [mailto:owner-ietf-ltans@xxxxxxxxxxxx]
> >>On Behalf Of Tobias Gondrom
> >>Sent: Tuesday, October 17, 2006 10:19 PM
> >>To: Tilo Kienitz
> >>Cc: ietf-ltans@xxxxxxx
> >>Subject: RE: next (hopefully final) revision (08) of ERS
> >>Hi Tilo,
> >>actually I did not intend to ignore your comment. I just thought that
> >>the answer Robert gave to your comment one day later was pretty good
> >>explaining why it does not make any real sense to sort the hashvalues
> > by
> >>date and not in binary order. (especially as I did not see any further
> >>answer from you on that)
> >>And I and the other authors fully agree with Robert's analysis and
> >>If Robert or I am wrong please let me know.
> >>Ps.: maybe there is also a misunderstanding about this issue, so if
> > you
> >>like to discuss this on the phone for a few minutes maybe we can
> >>exchange some arguments and see whether I missed s.th. or we can
> > report
> >>to the mailing list that the question could be resolved or we need
> >>further dispute on that. (if you like you can give me a call at
> >>>-----Original Message-----
> >>>From: Tilo Kienitz [mailto:tk-tlslist@xxxxxxxxxxxxxx]
> >>>Sent: Tuesday, October 17, 2006 11:39 AM
> >>>To: Tobias Gondrom
> >>>Cc: ietf-ltans@xxxxxxx
> >>>Subject: Re: next (hopefully final) revision (08) of ERS
> >>>Hello Tobias,
> >>>if this is the final revision, then I assume that you do not intend
> >>>to answer on my e-mail from 16 June regarding the sorting of the
> >>>hash values?
> >>>Kind regards
> >>>Tilo Kienitz