[LEAPSECS] stale leap second information

Zefram zefram at fysh.org
Sat Jan 31 11:10:28 EST 2015


Rob Seaman wrote:
>Any implementation should support the entire table of leap seconds.

I agree.  I'd arrange the DNS mechanism such that the queried domain
name encodes the month, and the resource record just represents the
TAI-UTC difference that applies that month.  Leap events are implied by
the difference changing between months.  NXDOMAIN for a particular month
means that the TAI-UTC difference for that month hasn't been decided yet.
TTLs for filled months can be arbitrarily large.  The effective TTL for
an NXDOMAIN response mustn't exceed a few months, and may be usefully
manipulated in anticipation of Bulletin C's release.

    $origin tai-utc.example.
    $ttl 100000000
    @ in soa . . ( 2015013100 100000 10000 1000000 100000 )
    1972-01 txt "10"
    1972-02 txt "10"
    1972-03 txt "10"
    1972-04 txt "10"
    1972-05 txt "10"
    1972-06 txt "10"
    1972-07 txt "11"
    1972-08 txt "11"
    ...
    2015-05 txt "35"
    2015-06 txt "35"
    2015-07 txt "36"
    2015-08 txt "36"
    2015-09 txt "36"
    2015-10 txt "36"
    2015-11 txt "36"
    2015-12 txt "36"

The TAI-UTC difference may also be encoded in A and AAAA records to
support lookup via gethostbyname().

-zefram


More information about the LEAPSECS mailing list