Re: Fwd: lsclocks and tai

Thomas Weißschuh <[email protected]> Thu, 13 Nov 2025 16:51:37 +0100
Newsgroups org.kernel.vger.util-linux
Message-ID <[email protected]>
Hi Sami,

On 2025-11-04 13:58:35+0000, Sami Kerola wrote:
> Meanwhile. There was a workplace discussion about TAI and I noticed
> unexpected lsclock.
> 
> $ lsclocks -o ISO_TIME,NAME | grep tai && TZ=right/UTC date --iso=ns
> 2025-11-04T13:58:14.176151344+00:00 tai
> 2025-11-04T13:57:47,177638531+00:00
> 
> One would hope above timings always match. If I force the timezone for
> the lsclock output will match, but this has downside of all the other
> listings use TAI zone.

I can't really follow what you mean with this. The only difference is
the printed time and that is expected, as TAI is (currently) 37 seconds
behind UTC.

> $ TZ=right/UTC lsclocks -o ISO_TIME,NAME | grep tai && TZ=right/UTC
> date --iso=ns
> 
> Perhaps the realtime outputs to use local time, and TAI to be always
> right/UTC. Any thoughts?

TAI does not have any timezone at all. Realtime does have one. For PTP
and RTC clocks it may depend on circumstances. Currently everything uses
+00:00 for consistency.


Thomas