On Tue, 05 Nov 2019 10:00:55 +1100 Greg 'groggy' Lehey <grog@lemis.com> wrote:
>
> [Moved to COFF]
>
> On Monday, 4 November 2019 at 16:59:22 -0500, John P. Linderman wrote:
> > I wrote a near-trivial "timestamp" command to make it easier to do time
> > arithmetic
> >
> > TZ=udt timestamp
> > 119 11 04 21 50 06 18204 1572904206 Mon Nov 4 21:50:06 2019
> > TZ=udt timestamp 0
> > 70 01 01 00 00 00 0 0 Thu Jan 1 00:00:00 1970
> > ...
>
> FreeBSD has this functionality in date(1):
>
> TZ=UTC date -r 500000000
> Tue 5 Nov 1985 00:53:20 UTC
> TZ=UTC date -r 1500000000
> Fri 14 Jul 2017 02:40:00 UTC
Shouldn't that be
Tue 5 Nov 1985 00:52:57 UTC
Fri 14 Jul 2017 02:39:23 UTC
given that this is UTC (which is 37 seconds behind TAI)?
The delta was closer to 18 in 1985 and 36 on 2017. But that doesn't matter.
No. POSIX says leap seconds do not exist for the computation of these numbers. Greg's numbers are POSIXly correct.
Warner