Leap Second in a graph | Hacker News Reader

Leap Second in a graph

twitter.com

24 pointsvyodaiken9 comments

One day, developers will rise up and thump the astronomers that create leap seconds, create a new "Astronomy Celestrial Time" zome, and try to forget about the disruptions this causes every couple years.

"Abolishing the leap second all those centuries ago was the best thing we'd ever done."

"I agree; now, come on, it's 3am - the sun's going to set soon, and if we want to enjoy this warm January day, we'd better be off."

Aren't climates, in general, changing anyway. Not to mention that for half of the globe that is an accepted fact anyway. So, temperature just isn't that compelling of an argument for this.

My joke was actually about the calendar itself precessing, since it would become decoupled from the Earth's rotation around the sun, but if you'd like to contradict the entire scientific establishment, you're welcome to.

I wasn't contradicting any establishment. I just don't think weather is a concern. Really, neither are dates. Our calendar system isn't that old, after all.

Edit to add: I'm also being over brief, since I'm mainly posting from my phone. I make no claim that there are reasons to ditch leap seconds. Or that I fully understand why we have them. I just personally feel that calendar weather isn't that compelling. The changes would be so minor per generation, I would wager most wouldn't notice it.

Now, let none of this get confused with climate change. Which I do feel we are contributing to and we should seek to control.

> create a new "Astronomy Celestrial Time" zome

TAI already exists and it's free from leap seconds. Operating systems and time protocols should absolutely use this. Too bad that both POSIX time and NTP time are infected by leap seconds.

What are the axes?

This is what it looks like to me:

X axis - time of day

Y axis - NTP server's delta from actual time

What you're seeing with the yellow lines is "leap smearing" by Google's public NTP time servers. They slowly diverge from the real time for the 10 hours before and after the leap second, so that they can have 60 instead of 61 seconds at midnight. See: https://cloudplatform.googleblog.com/2016/11/making-every-le....

Correct. The X axis is our estimate of actual time from GPS time. The Y axis is the deviation from estimate. The estimate includes a slew of about 5 minutes around the leap second.

Reply on news.ycombinator.com