[LEAPSECS] Metrologia on time

[Steve Allen]

On Tue 2011-08-02T10:53:56 -0700, Rob Seaman hath writ:

> PHK and I are on the same page with this.

> By all means announce leap seconds with significantly more notice.


Even so, I think it's important for operating systems that the kernel
never have to deal with any sort of "unusual" event.  In this case I
would say that "unusual" does not include leap years, for they are
predicted before the manufacture of any device and not subject to
change within the lifetime of the device.

I note that the Chinese contribution to Metrologia suggests a notion
akin to "leap centuries" which seems to indicate that all the earth
rotation anomalies would be adjusted by resetting the broadcast time
scale once every hundred years.

I'm not convinced that resetting the broadcast (and kernel) time scale
can ever be acceptable.  Any such thing begs the question of how to
encode tables of the offsets such that past civil times can be
reconstructed, but the whole point of tz/zoneinfo is to provide that
kind of tables.  Currently we see that ftp://elsie.nci.nih.gov/pub/
has tzdata2011h.tar.gz , so the politicians and bureaucrats have
already messed with civil time on 8 occasions during this year.

I'd much rather see a 3-year prediction of the content of the zoneinfo
"leapseconds" file while the underlying broadcast and kernel timescale
always remains uniform.  That allows for separation of the policies
and responsibilities of the broadcast time scale and the civil time
scale so that each one can adopt rules best suited for its purposes.

--
Steve Allen                 <sla at ucolick.org>                WGS-84 (GPS)
UCO/Lick Observatory--ISB   Natural Sciences II, Room 165    Lat  +36.99855
1156 High Street            Voice: +1 831 459 3046           Lng -122.06015
Santa Cruz, CA 95064        http://www.ucolick.org/~sla/     Hgt +250 m