[ntp:questions] Re: a new NTP implementation (dntp)

David L. Mills mills at udel.edu
Mon Apr 25 15:52:43 UTC 2005


Christopher,

Look more carefully. NTPv3 (rfc1305) was and is only a draft standard, 
not a full standard. That is because I refused then and now to package 
the document in Postel ASCII. My grant supporting NTPv4 specification 
now in progress specifically states that my contribution will not 
require Postel ASCII. All other standards bodies known to me prefer 
standards publication in PDF, which is the only format I will use. It 
might well be the case that a NTPv4 specification will never become a 
full standard unless somebody else does the conversion. Thus, there is 
no excuse to avoid deploying NTPv4 on the basis of standard.

Dave

Christopher Browne wrote:
> Oops! "David L. Mills" <mills at udel.edu> was seen spray-painting on a wall:
> 
>>David,
>>
>>Thanks for the spy. I took a look at the pages you mention. There's a
>>couple of points that should be made and they have nothing to do with
>>the excellence of lack of it with either their implementation our
>>"ours".
>>
>>1. I don't know why they even consider xntpd. That hasn't been the
>>standard of comparison for at least eight years.
>>
>>2. The IETF folks and especially me would very much welcome new kids
>>on the block, but only if they conform to spec, whatever that means
>>now. They can claim RFC-1305 and/or RFC-2030 or, best yet, they can
>>participate in the current IETF task force on NTPv4
>>specification. They would then get the chance to argue changes and
>>improvements, as was said about the frequency problem.
> 
> 
> Is any change likely on the NTPv4 standards?
> 
> That it isn't yet a draft means that the implementation also has no
> formal standing, which is probably why we can't get IBM Global
> Services to accept anything newer than xntp for use.  xntp is an
> implementation of something that _was_ accepted as a standard, whereas
> the same isn't true for NTPv4.
> 
> NTPv4 may be way better, but there are places where I can't get it
> deployed because there's no standard!



More information about the questions mailing list