[LEAPSECS] What is GMT?

[mike cook]

Le 05/01/2012 06:36, Harlan Stenn a écrit :

> ntpd can easily track SI seconds or "angle time" seconds.  The

> differences are small enough over a day to be easily amortized.

>

> It would not be all that difficult to create an NTPv5 protocol (for

> example) that would include "timescale" as a parameter.  There may be a

> way to do this using the v4 protocol, but I haven't thought about it at

> all.

>

> But that's pretty easy - there are other issues afoot here, like

> regardless of how one syncs time between machines, if one collects a

> timestamp on a box, what timescale is that timestamp in?


NTP4 already has extension field capability which could be used to  
identify and define the alternate scale data.
I was thinking along the lines that the reference, originate, receive 
and transmit timestamps be kept as now ( UTC in whatever form) so that 
if no extension in the packet, nothing to do. The transmitted scale(s) 
would  be piggybacked in  extensions  as an offset / frequency change to 
get to next alternate timescale tick, and on a per client basis. Up to 
the admin via ntp.conf to say what he wants to use . All managed scales 
would be transmitted. As there is usually only one hardware TSC/timer 
and RTC pair in a machine , those would need to be occupied by the 
chosen scale and served packets have their UTC fields reverse mapped 
from the TSC/timer data and managed alternate scale parameters. I guess, 
multiple alternate scales could be managed if the UDP packet space allows.

> What about storing these timestamps?  Displaying their values?

> Comparing them with other timestamps from possibly different timescales?

ntp management utilities would need  new options to list  display  and 
compare alternate timescale data.


>

> H

> _______________________________________________

> LEAPSECS mailing list

> LEAPSECS at leapsecond.com

> http://six.pairlist.net/mailman/listinfo/leapsecs

>

>