[ntp:questions] Re: Contradicting RFC<---> Implementation

David L. Mills mills at udel.edu
Sat Oct 8 16:21:28 UTC 2005


Any discrepancy between the NTPv3 specification (rfc1305) and the 
implementation (xntpd) must be accepted as-is and no repairs are 
anticipated. The developer corps does not maintain the NTPv3 
specification or implementation.

You question as it applies to the NTPv4 implementation is addressed at 
the NTP project page www.eecis.udel.edu/~mills/ntp.html. See the 
architecture briefing and note the error budget and sanity checks.


Remo wrote:
> After going down the rabbit hole,  I learnt that w32time is very poor.
>  And now my problem jots down to a single question:  In versions of
> xntp, the sanity checks of the clock-selection procedure shows that the
> (peer->dispersion is compared with NTP_MAXDISTANCE).  But the RFC
> suggests of using peer->synchdistance. Why is this =><= ??  I found
> that peer->dispersion is a subset in composing
> peer->synchdistance...but using synchdistance according to RFC could
> make the clock selection more strict right ?
> Please correct me if I am wrong and let me know why there is a
> difference in implementing.  I could not find the BugID that made this
> :o(
> Thanks for your time

More information about the questions mailing list