Timeline for GNU `make`, modification times, leap seconds and NTP
Current License: CC BY-SA 4.0
17 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Dec 17, 2024 at 21:36 | comment | added | Tomek Czajka | Sorry but your other answer clearly is wrong in light of the fact that Unix time jumps by 1 second during a leap second, as described on Wikipedia, and as I just confirmed by experiment. | |
| Dec 17, 2024 at 21:34 | comment | added | Marcus Müller | have you really read the other answer? I ask that because this confusion between "decomposed UNIX time" and "UTC" in the presence in leap seconds is explicitly addressed in there. | |
| Dec 17, 2024 at 21:31 | comment | added | Tomek Czajka | A simple experiment confirms for me that you are definitely wrong: time.is/Unix_time shows that Unix time modulo 60 matches the current time in seconds in UTC. Python's time.time() function also matches that. | |
| Dec 17, 2024 at 21:24 | comment | added | Marcus Müller | you might want to read the other answer on the question I linked to. If wikipedia contradicts that, wikipedia is wrong (or your interpretation of it). | |
| Dec 17, 2024 at 21:22 | comment | added | Tomek Czajka | Well the Wikipedia link I posted seems to disagree with you, it even has a table demonstrating an example of Unix time decreasing during leap second, so either your answer is incorrect or Wikipedia is very wrong about Unix time. | |
| Dec 17, 2024 at 21:18 | comment | added | Marcus Müller | no, it really doesn't. The counter is monotonous, it increases by 1 per second. That's the definition. | |
| Dec 17, 2024 at 21:17 | comment | added | Tomek Czajka | Hmm but if the reference point gets shifted forwards, the counter is going to decrease, which contradicts your claim it never decreases. | |
| Dec 17, 2024 at 21:16 | comment | added | Marcus Müller | the counter increases once per second, full stop; Leap seconds shift the reference point (epoch isn't anymore necessarily at exactly 00:00 on the first of January 1970). | |
| Dec 17, 2024 at 21:15 | comment | added | Marcus Müller | unix.stackexchange.com/questions/758932/… | |
| Dec 17, 2024 at 21:13 | comment | added | Tomek Czajka | I believe that is false. en.wikipedia.org/wiki/Unix_time "It measures time by the number of non-leap seconds that have elapsed since 00:00:00 UTC on 1 January 1970" | |
| Dec 17, 2024 at 21:12 | comment | added | Marcus Müller | unix timestamps are a counter that keeps increasing by 1 every second. Whether some humans declare that second a leap second or not doesn't matter to UNIX time. | |
| Dec 17, 2024 at 21:11 | comment | added | Tomek Czajka | But we're not talking about physics, we're talking about Unix timestamps... Sorry if I'm being obtuse but talking about time in physics doesn't resolve the question. Is the answer that timestamps in the filesystem are stored as a TAI offset rather than a UTC offset? | |
| Dec 17, 2024 at 21:10 | comment | added | Marcus Müller | They are not a problem at all, as they don't affect the time that has passed since epoch in any way differently than any other second. Leap seconds are a "converting time stamps to human-readable time" issue, not a time issue. Physics don't care. | |
| Dec 17, 2024 at 21:09 | comment | added | Tomek Czajka | OK so that answers the part about NTP, what about leap seconds (if there are more)? Are timestamps in TAI rather than in Unix time? | |
| Dec 17, 2024 at 21:08 | comment | added | Marcus Müller | and I answered that: still monotonous; leap seconds don't affect how many seconds have happened since some point in time. They affect how that number of passed seconds is converted to a human-readable datetime. . (unless your NTP client does things it shouldn't do. Modern NTP clients by default are safe, in that they never set the clock back, they just "slow down" time, so that timestamps remain monotonous) | |
| Dec 17, 2024 at 21:06 | comment | added | Tomek Czajka | I know the timestamps are usually monotonically increasing, but the question is about the possible exceptions: NTP adjustments and leap seconds. | |
| Dec 17, 2024 at 20:58 | history | answered | Marcus Müller | CC BY-SA 4.0 |