Leap seconds were introduced in 1972. The time_t that you are thinking of was invented somewhere around 1974, according to Unix historians.

Ironically, your kernel measures TAI time perfectly happily if unmolested, as that always ticks once for every SI second (ignoring oscillator drift). No slewing. No smearing. No stepping.

Isn't a second just how long light takes to travel 299,792,458 m?

Also, "since Unix time was created before leap seconds." I found this interesting for the simple reason I've never spent any time thinking about Epoch vs Leap Second histories.

Yes, but the point is that the UNIX timestamp doesn't count the the number of seconds elapsed since January 1, 1970. It counts 86400 seconds per day, regardless of how many seconds the day actually has (which can vary due to leap seconds).

Is this accurate though? Isn’t the number of seconds since 1970 absolute, and it’s up to the library generating the Gregorian date to take into account leap seconds? I suppose if these libraries are not taking into account leap seconds, then the actual rollover will be a few seconds earlier (or later?) than what we think.

See the definition of seconds since the epoch at https://pubs.opengroup.org/onlinepubs/9699919799.2018edition...

  tm_sec + tm_min*60 + tm_hour*3600 + tm_yday*86400 +
      (tm_year-70)*31536000 + ((tm_year-69)/4)*86400 -
      ((tm_year-1)/100)*86400 + ((tm_year+299)/400)*86400

Your time_t isn't the true number of SI seconds since the Epoch. It's that number, mucked around 27 times so far. Instead of leaving their kernels to count one tick per SI second (ignoring oscillator drift) many people slew, step, or smear their seconds count every time that there is a leap second.

Go and read about "right" versus "posix" timezones.

* https://unix.stackexchange.com/a/294715/5132

IIUC unix timestamp simply counts seconds (well defined interval of time) from epoch onwards. It is then up to different calendars to interpret this number as a given day / hour /... This is the beauty of it - it is always an accurate (well, at low speeds at least ;) ) measure of time. In other words, no, some days might not last 86400 seconds.

This is not correct. UNIX timestamps always have 86400 seconds per day, and consequently, they do not actually count the number of seconds since the epoch.

I stand corrected - thank you for the clarification!

And meter itself is defined as the length of the path traveled by light in vacuum during a time interval of 1/299,792,458 of a second.

So it's cyclic and doesn't make sense outside the earth reference frame

Edit: since 2019, 1 second is defined by taking the fixed numerical value of the caesium frequency ∆νCs, the unperturbed ground-state hyperfine transition frequency of the caesium-133 atom, to be 9192631770 when expressed in the unit Hz, which is equal to s−1

So a second makes sense in the galactic scale at least as long as Caesium-133 is a stable isotope in that environment

> Since 1967, the second has been defined as exactly "the duration of 9,192,631,770 periods of the radiation corresponding to the transition between the two hyperfine levels of the ground state of the caesium-133 atom

> realisation of the metre is usually delineated (not defined) today in labs as 1579800.762042(33) wavelengths of helium-neon laser light in a vacuum

No. A second is "The duration of 9192631770 periods of the radiation corresponding to the transition between the two hyperfine levels of the ground state of the caesium-133 atom."

No, that's how the metre is defined. The second is defined by the transition between two hyperfine states of cesium 133 (9,192,631,770 'oscillations').

Or lets say we colonize a planet with a significant time dilation (ie near a supermassive blackhole), how do you deal with this? Who has the canonical time?

AFAIK, there is no canonical time in relativity. There is the "local time", which is what your wristwatch shows, and the "time in place X", where X can be Earth, Mars, or some spaceship. These times might not be in sync, and they might even be distorted. Because, if place X is moving relative to you at a significant percentage of lightspeed, their seconds will be longer. Also, the further X is, the blurrier the concept of simultaneity becomes. Which makes the question "what time is it _now_ in place X" moot :)

If relativity don’t allow time travel doesn’t it make time canonical? A clock on Earth from telescope might look jumpy but only in forward direction after compensating for distance I think

Looking at a clock from Earth through a telescope isn't as straightforward as you'd think. The image you see of that clock is actually light (photons) emitted from Earth, which will take a while to reach you - like, 1 year, if you're 1 lightyear away. During that year, Earth has moved on, maybe blew itself apart. But you can't even tell, because information can't reach you faster than light :) So you can only see Earth's past, not Earth's "now". The further you are, the more "now" loses meaning.

Yeah but can it ever go back, other than by backing off faster than the speed of light? If it can’t then it’s at least monotonic, even if it’s not rate constant

All observers observer all clocks to advance monotonically forward in time, regardless of their location or relative speed, unless the clock is moving at the speed of light. But the rate that each clock is observed to advance at depends on things like relative speed, acceleration, and the curvature of spacetime.

... unless it's the system clock in Linux, with the hardware RTC set to be interpreted as local time, and one is watching a system boot on a machine that is set up as east of the Prime Meridian (i.e. the hardware RTC is ahead of UTC). (-:

* https://unix.stackexchange.com/a/600490/5132

Given that Unix time specifically doesn’t handle leap seconds, a system clock is not a clock by the conventional definition.

We'll probably come up with something similar to time zones once we become an interplanetary species.

"Local seconds since 1970", vs. "Static seconds since 1970".

You don't have to go to science fiction for this. Science fact will do. On Earth, with Atomic Time, we already have to account for general relativistic effects. Doco from the BIPM (the SI Brochure) already talks about accounting for them across the sizes of the measuring devices.

You'd just use some math to remove the influence of the black hole. Or for a more hands-on answer, you could have some reference beacons that are significantly further away from the black hole.

In a broad sense, the answer goes like this: Almost everything in the galaxy shares an approximate reference frame when it comes to velocity and special relativity. For general relativity and gravity, that's obviously a distortion so it doesn't count.

In the long term, to prevent very minor drift, we can use quasars as a reference.

Why would someone on Mars, or an asteroid, or orbiting venus, care about leap seconds on earth?

Number of seconds since a given time seems fine until we're travelling significantly outside of a common reference plane.

That’s exactly why op said we should abandon Unix time. The “who cares about leap seconds” standards is called TAI, and the difference between TAI and Unix time changes every time a leap second is introduced.

Unix time decrees every day has exactly 86400 seconds. But the introduction of leap seconds means that While most days have 86400, some earth days have 86401 seconds and potentially there could also be days with 86399

TAI makes sense everywhere, Unix time only on earth.

Because that's not how Unix time is defined. The leap second hack is hardcoded into it:

Unix time is [...] the number of seconds that have elapsed since the Unix epoch, minus leap seconds.

- Wikipedia

If you are on Mars you'll have to update your computer time every ~6 months when Earth release a new table of leap seconds caused by e.g. earthquakes.

Or we just pretend they never happened.