On 10/17/2026 1:38 PM, Arthur Olson via tz wrote:
The bad news is that leap hours are best handled with a new version of
time zone binary files.
The good news is that there are centuries to get the work done.
The challenge I see for leap hours is figuring out the base time that
one is using as a guide.
tzdb defines offsets from UTC. UTC is the standard. Theoretically we can
(and do) have zones defined as being minutes offset from UTC.
Leap seconds change UTC. We can look through a file and see where the 27
seconds were added. If the scheme continued and negative leap seconds
were allowed we would be able to look in a file to see where seconds
were removed or additional seconds were added. All on a global scale.
A leap hour would need to be defined similar to leap days or the
Gregorian calendar change ... if someone is counting the number of
hours/minutes/second since an epoch they would need to know that UTC
skipped an hour on a certain day (and be happy that it was not a "fall
back" where UTC repeated itself for an hour).
But, as you say, this is a problem for future generations. Generations
that may not be tied to "Earth time" on a universal basis.