[tz] leap_second.list not updated after latest IERS Bulletin C

Brooks Harris brooks at edlmax.com
Fri Dec 8 19:58:44 UTC 2023

On 12/7/2023 9:29 PM, Paul Eggert via tz wrote:
> On 12/7/23 13:21, Geert Hendrickx wrote:
>> NTPsec logs the following at startup:
>> CLOCK: leapsecond file ('/etc/ntp/leap-seconds.list'): good hash 
>> signature
>> CLOCK: leapsecond file ('/etc/ntp/leap-seconds.list'): loaded, 
>> expire=2024-06-28T00:00Z last=2017-01-01T00:00Z ofs=37
>> So I *guess* (without having looked at the code) that it actually cares
>> about the expiration date as well.
> I think you're right NTPsec does care, though I vaguely recall that if 
> the file has expired the only issue is an unwanted log message.
> If people are running NTPsec and configuring it to use TZDB's 
> leap-seconds.list, that unwanted log message could be an issue. I just 
> now checked the NTPsec source code, though, and by default it uses 
> this URL:
> https://www.ietf.org/timezones/data/leap-seconds.list
> which hasn't worked in a while; the contents are simply "ietf.org is 
> no longer serving this file." So it may be that we don't need to issue 
> a new TZDB release merely because 2023c's leap-seconds.list will be 
> out-of-date soon.
>> (but we periodically download it from IERS, not from TZ)
> In that case your setup is OK as-is, though we may still need to hear 
> from other people to see whether they're relying on the TZDB copy of 
> leap-seconds.list.
> TZDB uses the NIST version of leap-seconds.list rather than the IERS 
> version, as the NIST version is clearly public domain and so this way 
> we don't have to worry about copyright issues. However, the IERS 
> version should work fine with either NTPsec or with other downstream 
> uses, such as TZDB itself (that is, if you're not worried about 
> copyright).

In my opinion TzDb should update the leap-seconds.list date and issue a 
TzDb release because there's no telling who might be using it and 
perhaps relying on the expiry date.

The current expiry date at 
https://ftp.iana.org/tz/tzdb-2023c/leap-seconds.list is
#    File expires on:  28 December 2023

The current expiry date at 
ftp://ftp.boulder.nist.gov/pub/time/leap-seconds.list is
#    File expires on:  28 June 2024

There are other sources at IERS besides Bulletin C. In my development 
work I've been using:

This file has a different form, using MJD rather than "number of seconds 
since 1 January 1900, 00:00:00".

Note that, as I understand it, Bulletin  C is the only official 
"product" of the IERS. But this Leap_Second_History.dat file has been 
being issued and maintained for years.

I cannot comment expertly on "copyright", but it seems to me anything 
from IERS must be public domain, isn't it? How is it not?


More information about the tz mailing list