[ntp:questions] Re: Time set backward

David L. Mills mills at udel.edu
Sat Jan 7 13:57:11 UTC 2006


Garrett,

I don't have much patience with the standards process that tends to 
create a reality not sustainable in practice. If POSIX has no way to 
discipline the clock frequency, the clock frequency is not disciplined 
and the platform not reliably synchronized. Same thing happened with 
Digital DTSS and for that matter the IBM 9037 Sysplex Timer. The reality 
is that standards monkeys are neither engineers nor physicists and don't 
understand the physical basis of the technology.

Dave

Garrett Wollman wrote:engineering
> In article <dpmsdo$7q7$1 at dewey.udel.edu>,
> David L. Mills <mills at udel.edu> wrote:
> 
> 
>>Thanks for the clarification. How do I discipline the frequency of the 
>>CLOCK_REALTIME clock?
> 
> 
> Administrative operation outside the scope of the standard.  I'm
> actually surprised even clock_settime() slipped through -- I can only
> suppose that it was allowed because the committee felt it important to
> define how timers were supposed to behave when the system clock was
> changed.
> 
> The interface NTP uses, if sufficiently standardized across platforms,
> could conceivably be standardized, but I doubt there would be much
> interest in doing it as a part of the POSIX process.  (The deadline
> for work items in the next revision of 1003.1 is Sunday, by the way.)
> 
> -GAWollman
> 




More information about the questions mailing list