[Ntp] Antw: [EXT] Re: CLOCK_TAI (was NTPv5: big picture)

[Ulrich Windl <[e-mail]>]

>>> Philip Prindeville <[e-mail]> schrieb am 05.01.2021
um
20:56 in Nachricht
<[e-mail]>:

> 
>> On Jan 5, 2021, at 8:51 AM, Magnus Danielson <[e-mail]> wrote:
>> 
>> So I think we need to choose between two different scenarios, either one
>> where we allow multiple timescales being sent, without TAI-UTC
>> difference at all times, or we go for a unified timescale meeting the
>> ideal model properties, and say that we need to be able to resolve the
>> issue. There is no real way around that central question.
> 
> 
> It might help to stop conflating issues.
> 
> Clock synchronization is a simple problem requiring linear, non-leaping 
> time.
> 
> Calendaring is an orthogonal issue of breaking groups of seconds into 
> human-friendly units of minutes, hours, days, years, etc.
> 
> NTPv4 tried to resolve both issues because it seemed convenient at the time:

> in retrospect, it’s not clear that it didn’t create more problems.
> 
> I say that we drop the unnecessary encumbrances, and shift attention to the

> necessities that have since emerged that NTPv4 didn’t address (like clarity,

> good security, better extensibility, etc).

But don't forget: Even if NTP is able to keep perfect time within it's
processes, it's of little use if the operating systems have no access to that
"perfect time".
And AFAIK "struct timespec" is the best UNIX-like operating systems have; I'm
unsure about the MS-Windows world. So we need a solution to fill the gaps
between NTP and OS.

> 
> -Philip
> 
> _______________________________________________
> ntp mailing list
> [e-mail] 
> https://www.ietf.org/mailman/listinfo/ntp