Re: [Ntp] Local time in NTPv5?

[Danny Mayer <[e-mail]>]

On 7/27/22 11:01 AM, Miroslav Lichvar wrote:
> There are some very simple devices like clocks or security cameras
> that use NTP for synchronization, but they need to know the time zone
> in order to display local time. Selecting the time zone and/or
> updating the data may be difficult. I have heard of some cases where
> people gave up on the configuration and instead patched their NTP
> server to compensate for the offset between the default time zone and
> local time.
>
> I'm wondering if NTPv5 could better support this use case. In my draft
> there already is UTC, TAI, UT1, and leap-smeared UTC. Local time could
> be the 5th timescale to be requested by the client. The timescale
> offset field could be the offset to UTC in minutes (in seconds it
> would not fit the 16-bit field for some zones).
>
> The assumption is that the client doesn't know where it is, but it
> expects the server (e.g. from the DHCP NTP option) to be in the same
> time zone. The software would be simplified and no time-specific
> configuration is required.
>
> Does that seem reasonable?
>
This makes no sense at all. NTP only knows about UTC. All major 
operating systems store timestamps in UTC. Time-zones are just a display 
issue. The software displaying the timestamps is responsible for 
displaying timestamps in the appropriate way. You are incorrect in 
stating that no time-specific confguration is required. NTP doesn't know 
where a client is nor does it have any way of telling. The client does 
know. Don't expect NTP to deal with the Olson time-zone database.

Danny