3 ms·
It would be even simpler to use a standard without leap seconds (or at least just without negative leap seconds). The drift happens slowly enough that nobody wi
by hansvm 3y ago
It would be even simpler to use a standard without leap seconds (or at least just without negative leap seconds). The drift happens slowly enough that nobody will notice the qualitative elements of 8am changing slightly over their lifetime.
- layer8 3y agoThe drift is speeding up long-term though.
- hansvm 3y agoI'm optimistic we'll have better synchronization mechanisms in the distant future than "8am == sunrise, +/- a few hours". That said, interestingly, the quadratic nature of that problem (the earth slowing down progressively faster over time) means that the gap between when we have to care about my solution (because we accumulate too many leap seconds per lifetime) and the leap second solution (because you can only insert a little over 365 per year with the current protocol) is much smaller than the gap between now and when we would have to care about my solution. If you're not too picky, the two solutions "expire" around the same time. The max rate of leap second insertion only admits an hour of drift per decade, which is slow enough that I still think most people wouldn't care, and they might like if our time handling is simple enough that we can't repeat the recent xz supply chain issue in hundreds of thousands of lines of increasingly complex datetime nonsense.