Yeah the way Jiff's leap second support worked (before I ripped it out) is that it would just read from `/usr/share/zoneinfo/leap-seconds.list` (or `/usr/share/zoneinfo/leapseconds`, whichever was available) and use that with some rudimentary caching. That way, Jiff isn't responsible for keeping the list up-to-date. The sysadmin is. Just like what we do for tzdb.
Indeed. The problem is, in the absence of one good solution like that, there are no good solutions, as is the case with web JS. Currently (on the desktop or laptop), the best bet is hifitime's leap-seconds parser, but then the programmer is still responsible for coming up with a sane fallback path for all their target systems. (Which can be tricky for individuals to do, e.g., I have no way to tell whether the tz database is in an expected location on macOS or iOS.)
