On Fri, 1 Feb 2019 at 07:44, Tony Finch <dot@dotat.at> wrote:

Paul Eggert <eggert@cs.ucla.edu> wrote:
>
> One cannot simply translate the IERS file to the NIST file, as the NIST file
> has information that the IERS file lacks, namely, the last time that the data
> were changed.

Isn't that always 5ish months before the last leap second? :-)

Well, both have a #$ line, but the purpose is slightly different. For NIST, the value is the last time that only the data were changed (currently 2016-07-08T00:00:00Z), but for IERS it's the last time the file as a whole was updated, including the metadata (currently 2019-01-07T14:19:26Z).

I would imagine, yes, IERS wouldn't be keen on adopting NIST's practices for this line, but I don't necessarily see the reverse change being particularly disruptive, if we were to take IERS' file as-is. In the worst case, anyone relying on the less-often-updated NIST version of the line would just end up pulling the same, unchanged data once every six months.

--
Tim Parenti