Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Time is an illusion, Unix time doubly so

netmeister.org

Post reply on HN
Print view View original

81–90 of 104 posts

Re: Time is an illusion, Unix time doubly so

#81

Post by zokier »

gjulianm wrote: post #16

Maybe I am misunderstanding the post, but for me the beauty of Unix time is precisely that all the weirdness with dates is abstracted away to the "conversion code" so that you only deal with "seconds". Timezones, leap seconds... all of that only matters when you're showing the user a date. For recording and calculations, it doesn't.

UNIX timestamps are horrible. You can not do calculations with them, with fractional part they do not sort correctly, and you actually can not reliably convert them back and forth to iso timestamps. Basically to do anything useful, you need additional bit carried along the timestamp to tell if its leap second or not, which is extremely awkward and most APIs do not support that.

Re: Time is an illusion, Unix time doubly so

#82

Post by _kst_ »

mlichvar wrote: post #53

The article missed an opportunity to describe how spectacularly can things break when the 32-bit time_t overflows in Y2038. If you still have such a machine (preferably without any valuable data), try setting the date right before the overflow with this command date -s @$[2**31 - 10] and see if you can recover the system without reboot. I have seen different daemons stopped responding and just consuming CPU, NTP clie…

> date -s @$[2*31 - 10]

A digression: The documented bash syntax for arithmetic expansion is $(( EXPRESSION )) . I see that $[ EXPRESSION ] also works, but I don't see it documented anywhere. (Both syntaxes also work in zsh.)

Re: Time is an illusion, Unix time doubly so

#83

Post by chipsa »

jzl wrote: post #63
tonmoy wrote: post #43

Earlier quoted context omitted.

Why is it bad planning? I can’t think of a better alternate plan.

It’s bad planning because 64-bit time stamps were supposed to provide coverage for “22 times the expected lifetime of the universe”, but by being the only OS to use those 64 bits to count nanoseconds, they only bought a couple more bits over 32-bit timestamps. What they could have done differently is not used nanoseconds for the epoch counter. Milliseconds seem like they would have offered the perfect balance.

Windows uses 100 nanosecond resolution for NTFS time stamps.

Re: Time is an illusion, Unix time doubly so

#84

Post by somat »

state_less wrote: post #5

The article makes a passing reference to atomic clocks, which are fascinating. The folks over at MIT are working on improved clocks that can get below 100ms error over the the current lifespan of the universe. https://news.mit.edu/2022/quantum-time-reversal-physics-0714

CuriousMarc has an excellent episode on an atomic clock.

https://www.youtube.com/watch?v=eOti3kKWX-c

I like to pretend I can fix computers, this guy can actually fix computers, which is probably why I find his show so fascinating.

Re: Time is an illusion, Unix time doubly so

#85

Post by 8bitsrule »

Haven't astronomers already found a way to solve this problem? by using the standard Epoch.

"there has also been an astronomical tradition of retaining observations in just the form in which they were made, so that others can later correct the reductions to standard if that proves desirable, as has sometimes occurred." [https://en.wikipedia.org/wiki/Epoch_(astronomy)]

Re: Time is an illusion, Unix time doubly so

#86

Post by thaumasiotes »

mannykannot wrote: post #21

"On Unix systems we measure time as the number of seconds since "the epoch": 00:00:00 UTC on January 1st, 1970.... this definition is not based on something sensical such as, say, the objective frequency of vibration of a Cesium-133 atom, but on a convenient fraction of the time it takes a particular large rock to complete a full rotation around its own axis." Well, seconds have not been defined as "a convenient frac…

> Well, seconds have not been defined as "a convenient fraction of the time it takes a particular large rock to complete a full rotation around its own axis" for quite some time

Seconds have not ever been defined that way, because the time it takes for the earth to complete a full rotation around its own axis (the "sidereal day") was never a question of much interest. It's mostly relevant to astronomers.

Seconds were always defined in terms of the synodic day, the time it takes for a point on the earth that is aimed directly at the sun to rotate around the earth until it is once again aimed directly at the sun.

They still are defined that way, in the sense that the only purpose of other definitions is to match the traditional definition as exactly as possible. If cesium started vibrating faster or slower, we'd change the "official definition" of a second until it matched the traditional definition. Given that fact, which definition is real and which isn't?

Re: Time is an illusion, Unix time doubly so

#87

Post by macintux »

samatman wrote: post #34
_jal wrote: post #22

Earlier quoted context omitted.

Oh, if you own someone's clock, you can do all sorts of damage. Want to expire everyone's passwords? Blow up their credit card processing? Those are just the obvious attacks.

Yes but bringing down a database should not be among those things. Surely we can agree on that.

I’m curious whether there is any database that handles this scenario correctly…or even whether there is a correct behavior for a situation where the system clock is rewound.

Re: Time is an illusion, Unix time doubly so

#88

Post by pfarrell »

macintux wrote: post #87
samatman wrote: post #34

Earlier quoted context omitted.

Yes but bringing down a database should not be among those things. Surely we can agree on that.

I’m curious whether there is any database that handles this scenario correctly…or even whether there is a correct behavior for a situation where the system clock is rewound.

Good question. I used to test an auction app I maintained on SqlServer but I never pushed it more than a few hours into the future. To be clear: my experience with MongoDb was a one off local issue. No idea if there are server protections etc.

Re: Time is an illusion, Unix time doubly so

#89

Post by mannykannot »

thaumasiotes wrote: post #86
mannykannot wrote: post #21

"On Unix systems we measure time as the number of seconds since "the epoch": 00:00:00 UTC on January 1st, 1970.... this definition is not based on something sensical such as, say, the objective frequency of vibration of a Cesium-133 atom, but on a convenient fraction of the time it takes a particular large rock to complete a full rotation around its own axis." Well, seconds have not been defined as "a convenient frac…

> Well, seconds have not been defined as "a convenient fraction of the time it takes a particular large rock to complete a full rotation around its own axis" for quite some time Seconds have not ever been defined that way, because the time it takes for the earth to complete a full rotation around its own axis (the "sidereal day") was never a question of much interest. It's mostly relevant to astronomers. Seconds were…

That's an interesting point of view, but is it not the case that the length of the synodic day varies throughout the year, on account of Kepler's 2nd. law? - IIRC, this is the first-order effect in explaining the analemma. The length of the sidereal day is also variable, but more slowly (at least if we put aside atmospheric and earthquake effects, which I guess affect both equally.)

I think we need both a way to reckon our everyday experience of passing time and a constant measure, and one cannot be just a special case of the other. Personally, I feel that introducing leap seconds into Unix time just muddied the waters.

As to the sidereal day being of little interest outside of astronomy, James Watt's patron/partner Matthew Boulton built a sidereal clock and tried hard to find a purchaser (even having it shipped to Catherine the Great in hopes of sparking some interest) without success. It was on display in his Birmingham home when we visited a few years ago.

https://www.thefreelibrary.com/Boulton+clocks+on+to+the+past...

Re: Time is an illusion, Unix time doubly so

#90

Post by thaumasiotes »

mannykannot wrote: post #89
thaumasiotes wrote: post #86

Earlier quoted context omitted.

> Well, seconds have not been defined as "a convenient fraction of the time it takes a particular large rock to complete a full rotation around its own axis" for quite some time Seconds have not ever been defined that way, because the time it takes for the earth to complete a full rotation around its own axis (the "sidereal day") was never a question of much interest. It's mostly relevant to astronomers. Seconds were…

That's an interesting point of view, but is it not the case that the length of the synodic day varies throughout the year, on account of Kepler's 2nd. law? - IIRC, this is the first-order effect in explaining the analemma. The length of the sidereal day is also variable, but more slowly (at least if we put aside atmospheric and earthquake effects, which I guess affect both equally.) I think we need both a way to reck…

> but is it not the case that the length of the synodic day varies throughout the year, on account of Kepler's 2nd. law?

I don't see the relevance; the length of the sidereal day isn't constant either.

Variation in the length of the synodic day is the reason a day may contain other than 86400 seconds. If days are not of constant length, you could vary the length of a second on a day-to-day basis, or you could define the length of a "reference day" and then define the second as a convenient fraction of that. We have taken the second approach (though the first was used historically).

But we don't care about the vibration of cesium; if that were to change, we would adjust by changing the definition of a second, not by accepting that seconds were now of a different duration than before. Thus, the fact that cesium is referenced in an "official" definition of the duration of a second is meaningless. The officialness of that definition is illusory; in reality, seconds continue to be defined as a convenient fraction of an average day.

Post reply on HN