[ntpwg] updated ntpv4 draft available
David L. Mills
mills at udel.edu
Mon Feb 5 11:23:13 PST 2007
On re-review I found additional serious errors in the draft.
1. Figure numbering is inconsistent and wrong in places. Figures 1 and 4
are the same diagram.
2. The skeleton appendix is all messed up. Some tokens are joined, some
lines are missing and some are broken in the middle of a token. There is
a block of about 20 lines missing in the middle of a routine.
3. You guys did a commendable job in bashing the tables, which look just
fine, and the figures, but some figures seriously resist bash and are
really hard to decipher. One of the most important figures, that for the
poll process, is missing.
4. Some variable names are confusing and some are inconsistent. For
instance, the naming convention for upper case Greek is sometimes to
precede the lower case name with cap, other cases simply capitalize the
Greek name. I never did figure out where the variable varphi came from
or that the prefix var means. There are several misuses of the Greek
names phi and psi.
5. Reading carefully for content, there are several cases where the text
is just plain wrong. Some of that is my fault in the original, some of
it is lost in translation.
6. There are several places where figures and tables are broken between
pages, but that's a formatting issue, presumably easily fixed.
Well, that's the bad news. Now for what I hope is good news.
I spent the last four days carefully reading and revising the text,
repairing and verifying from the original the skeleton and revising the
variable names for consistency and correctness. I took out all the
Postel formatting, which would be useless anyway, removed the more
obnoxious figures and revised the text to emphasize the skeleton for
detail rather than the figures. Where figures really did help, I left
ALong the way, I renumbered the tables and figures and left a list of
each just before the document. I checked the table and figure referenced
in the text and corrected what broke. I assume you can stuff the lists
in at the right place and have your document processor do the formatting
and page numbering.
I read every sentence in the draft and considered whether it was
necessary, accurate or poorly written. I read the whole thing in part
and in whole several times. Just like I did for my book.
I pulled out all references to my book and put a note near the beginning
that additional information is available in the book, the UDel report
and the web site.
I have a few residual concerns.
1. Those considering SNTP viz. full NTP will need to read up through the
end of the on-wire section and they might not have the patience for
that. Perhaps mention of this should be early in the presentation.
2. There remain a few inconsistencies between the variable names defined
in the text and in those defined in the appendix. I am too exhausted to
fix that. In any case the text is consistent and the appendix is consistent.
3. While I fixed the little broken things with the tables, the figures
might be improved. I invite the graphic artists among us to help with that.
My expectation is that the text itself without the skeleton is
sufficient for most folks who only want to know how it works and not the
details. I read it through with this in mind and it works for me. The
bottom line is whether a skilled programmer can engulf the thing and be
able to implement a conforming implementation and verify with the
reference implementation. I invite discussion on this point.
The latest draft is attached. It is a bit smaller than before, since I
pulled out some extraneous material in the clock discipline section. It
is in the classic form of one sentence per paragraph, which should be
easy to format. May I suggest this as the basis for comments and markup
and save the formatting for the final draft?
If you wish, I can do the markup, either in response to general guidance
or specific typos. I encourage this, as the ancient text editor I use
(word5!) is eye-friendly for folks with dim vision and modern editors
are not. I'm not sure if diff works with long lines, but I do reserve
the right to fix broken things that I might find myself.
If/when we all agree, you or designate can do the magic reformatting and
distribute for last call. This can all happen by the deadline.
Brian Haberman wrote:
> Agree in principle. I would *strongly* encourage people to get
> comments to the editors or the mailing list by February 23. That will
> give the editors sufficient time to incorporate changes or engage people
> in discussions on issues before making changes prior to the draft deadline.
> Jim Martin wrote:
>>Thanks very much for starting the review. Your previous comments were
>>very helpful in cleaning things up quite a bit. From todays comments,it
>>seems there's more to be done, so I'd be happy to work with you in
>>whatever way is easiest.
>>As for the attribution, I think we'd be happy to reference your book as
>>a non-normative informational reference, and your earlier report should
>>indeed be mentioned in the acknowledgments.
>>On a more general note to all list members, I just want to continue to
>>stress the editing teams desire to finish up this document in a timely
>>manner. To that end, I again ask that people review and comment on the
>>-04 draft and allow us to come up with a solid -05 before the Prague cut
>>off. That means getting reviews back with specific, detailed comments
>>before February 26th, which would give the editors one week to integrate
>>the changes and make the draft submission deadline. Karen, Brian, do you
>>On Feb 1, 2007, at 11:00 AM, David L. Mills wrote:
>>>I've reviewed about a third of the document and made lots and lots of
>>>markups. There are misplaced tables, figure descripancies and figure
>>>errors. These are all fixable, but does call into question the last
>>>call status. This is definitely not in last call status. The most
>>>serious problems are inconsistent variable names.
>>>It would be a serious effort to list each and every markup; there are
>>>dozens and dozens of them. What I propose is to mark up a copy of the
>>>current document in ASCII and return it for diffing, further markup
>>>and reformatting. I can do the text part, but not the table and figure
>>>references and tables and figures themselves.
>>>I could simply send the paper markup, but even this is a lot more work
>>>than I would like. I would rather speed through the edit with my
>>>favorite eye-friendly editor and search and destroy inconsistencies as
>>>I find them.
>>>It would be polite to reference my earlier report as a source of much
>>>information included in the RFC. It would be of considerable help to a
>>>serious implementer to point out, maybe in a footnote, the detailed
>>>rationale and performance analysis for some of the more intricate
>>>algorithms can be found in my book and/or the NTP website.
>>> Odonoghue, Karen F CIV NSWCDD, W13 wrote:
>>>>From: Harlan Stenn <stenn at ntp.isc.org>
>>>>To: Brian Haberman <brian at innovationslab.net>
>>>>CC: Harlan Stenn <stenn at ntp.isc.org>; Odonoghue, Karen F CIV NSWCDD,
>>>>W13; ntpwg at ntp.isc.org <ntpwg at ntp.isc.org>
>>>>Sent: Wed Jan 31 21:48:04 2007
>>>>Subject: Re: [ntpwg] updated ntpv4 draft available
>>>>I've updated http://ntp.isc.org/IETF with the mailing list info.
>>>>ntpwg mailing list
>>>>ntpwg at support.ntp.org
>>>ntpwg mailing list
>>>ntpwg at lists.ntp.isc.org <mailto:ntpwg at support.ntp.org>
>>ntpwg mailing list
>>ntpwg at support.ntp.org
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
More information about the ntpwg