Hacker News Reader

Comment by cornholio | Hacker News Reader

It's basically the same thing, but harder to achieve technically because many things assume clocks are moving all the time. For example timers and timeouts are frequently implemented as loops polling system time has reached a certain future value. Many things break or malfunction if a large number of things happen at the same time, for example logging, crons etc. Imagine investigating a cybersecurity event perpetrated in the gap and finding all file dates and logtimes as unreliable.

View on HN

Those things should be on UTC anyway (even on earth). Timezones are just for displaying to humans.

Probably, but even Windows would break in this scenario. It interprets the hardware clock as being in local time and derives UTC from that (you can change that with a registry change, but it isn't very reliable in my experience)

Keeping the clock in UTC is reliable on Windows 10/11. Windows will happily read and write to the clock in UTC, and things like NTP sync just work.

Source: 2 years of heavy Windows 10/11 usage with that setting.

They really should have been on TAI. I guess it doesn't matter now if we won't get any more leap seconds.

It's worth noting that even atomic clocks run on Mars faster than on Earth, due to the different gravity. So you might expect say half of the 50 microsecond daily gain quoted in this article for the Moon, 10 miliseconds per year or a full second per century, more than enough to make you your spaceship completely miss its intended orbit. And martians will definitely want to run their own atomic clocks, making sync very problematic.

So TAI is not the end all for what I expect to be a very complicated technical problem.

Not sure whether sync will be 'very problematic'.

We already have that sort of sync happening today: when your computer is off, it runs a quartz watch. When you turn it on, it syncs via ntp.

Some special boffins on Mars would run atomic clocks, and worry about synchronisation with earth.

For the rest of the people running quartz clocks, they wouldn't even notice.

> Some special boffins on Mars would run atomic clocks, and worry about synchronisation with earth.

I think at some point, you have to redefine utc to be based on an estimated atomic clock at the center of the sun, because it's easier for mars and earth to estimate that, than for earth to go on its merry way and mars to estimate the time on earth. Relativism is fun?

> It's worth noting that even atomic clocks run on Mars faster than on Earth, due to the different gravity.

The biggest effect is the Sun's gravity, followed by orbital velocity. The relative masses of the planets is negligible by comparison, though in online discussions that's almost all anybody ever seems to talk about.

I can't find confirmation through other sources, and can't credibly validate it myself, but the best estimate I found for time dilation between two clocks on Earth and Mars is roughly 5.6 parts per billion, or ~177 milliseconds per year. See https://space.stackexchange.com/questions/33590/what-is-the-...

Wow, that's huge, and the Earth - Mars offset is likely perturbed by things like the motion of Jupiter etc. Keeping the time representations compatible and in sync will be a full time job for many people.

Timers sound like they should be using a monotonic timer actually, so they are not affected by the system clock being adjusted.

How is stopping the clock for 40 minutes human friendly? I'm a night owl and there are many things in my life that happen between 00:00 and 00:40, can't imagine pretending they all somehow happen in the same moment.

Reply on news.ycombinator.com