t-a-i

> It will indeed not, and if you care about that difference, you should not do that, and use something like https://github.com/qntm/t-a-i instead. That solution, as with most simple leap-second solutions, relies on the fixed file being kept up-to-date ad infinitum to account for any future leap seconds. Unfortunately, there isn't and still won't be any way to get a current list of leap seconds from JS, except for downloading them from someone else's API.

> Does this mean that converting two DateTimes into their POSIX time and subtracting them will not yield an accurate representation of number of seconds between two timestamps? It will indeed not, and if you care about that difference, you should not do that, and use something like https://github.com/qntm/t-a-i instead. > Am I naive? It seems like POXIS ignoring leap seconds is a major mistake. Depending on your needs it can be either seen as a massive mistake or a big convenience win. In either case it's arguably not a complaint with POSIX, but rather with the commercial and scientific world at large using UTC instead of TAI (and there are indeed people arguing for abolishing leap seconds!). POSIX in turn just followed that standard.