3 ms·
Weird, I came to a different conclusion after reading the article. There's already a graceful solution to non-monotonic time, which mitigates most of the probl
by klyrs 4y ago
Weird, I came to a different conclusion after reading the article. There's already a graceful solution to non-monotonic time, which mitigates most of the problems: smear, don't leap. Only, it's not a universal solution, so various systems are out of sync during the smear. Solution: petition for a standardized smearing strategy. But yeah, leave "leap" seconds in the past.
And, maybe, don't run sub-second benchmarks with a wallclock.
- daine 4y agoIf smearing were adopted as standard, the "seconds," "minutes," and "hours" appearing in timestamps would no longer correspond to literal seconds minutes, and hours of duration, even in principle. That seems very misleading and bad.
- ncmncm 4y agoSmearing is absolutely the worst of all possible choices. Instead of one second, you are out of sync with the whole world for all of 24 hours. And you are fooling with things for the whole period. Of course it was Google who picked the worst of all possible choices.
- joshuamorton 4y agoBut, its practical success does in many ways prove that mild inconsistencies between different time systems are...fine, and so a leap-less approach wouldn't cause any issues.
- klyrs 4y agoNo, the absolute worst possible choice is geoengineering -- literally moving mountains to correct the effective radius of the earth.
- ncmncm 4y agoThere is this word "possible" there. You might call the biggest earth-moving company in the world and see if they will give you a quote for that.