Talk:Unix time/Archives/2013

Is Unix time really an integer?

The section "Encoding time as a number" begins "Unix time is a single signed integer number...", but then proceeds to show "Unix time" values (like in the tables below) that are not integers. I realize a time_t is an integer, but is "Unix time" -- in the sense of time-keeping -- really an integer? My understanding is that Unix time is at heart a continuous variable, with discrete representations possible at various levels of precision -- integer, double, long double, etc. But I'm not a time geek, so maybe someone can enlighten me. — Preceding unsigned comment added by DKMell (talk • contribs) 18:09, 9 May 2013 (UTC)

ISO C defines type time_t to be an arithmetic type (only). POSIX and Unix have always defined time_t more specifically to be one of the signed integer types, usually int or long. The tables show values with fractional seconds to illustrate how leap seconds are handled by systems utilizing real-time clocks with sub-second precision; the actual time_t values are still implemented as integer types. — Loadmaster (talk) 00:05, 22 May 2013 (UTC)

And most UN*Xes also have calls that fill in structures containing Unix time (usually represented as time_t) and fractions of a section, such as gettimeofday(), which fills in a struct timeval that has tv_sec and tv_usec fields, the former being seconds since the Epoch and the latter being microseconds since that second. That's still not continuous, though, it just has finer resolution. Guy Harris (talk) 00:47, 22 May 2013 (UTC)

Year 292,277,026,596 is illegal as fact for encyclopedia

At 15:30:08 UTC on Sun, 04 December 292,277,026,596, 64-bit versions of the Unix time stamp will cease to work, as it will overflow the largest value that can be held in a signed 64-bit number.

I think, this is not a fact which can be stated in the encyclopedia. This year is something hypothetical for Mankind, Unix, or Solar system ([1]). Sun will be dead at this time and same should do Unix. I think there is no realiable source of this year problem; also calculation of this date is noncorrect. Length of year will change with time, and when there is no Earth, there is no legal year. And when there is no mankind, there is no calendar keeping. `a5b (talk) 22:52, 30 January 2012 (UTC)

The length of the solar year is irrelevant. Either UTC will be synchronized using leap seconds (see Coordinated Universal Time) or allowed to go out of sync. Neither will have an effect on The year 292,277,026,596 problem. The idea that "when there is no Earth, there is no legal year" is patent nonsense.

As for the argument "when there is no mankind, there is no calendar keeping", see Wikipedia:What_Wikipedia is not#Wikipedia is not a crystal ball. We have no idea how long humanity will last. All we can do is guess. It is possible that we will go extinct a month from now, yet we still write Wikipedia articles about future events past that. We cannot completely rule out the possibility that something (perhaps a Postbiological Civilization or even a Boltzmann brain with a retrocomputing hobby) will still be using 64-bit Unix Time in the year 292,277,026,596.

Also see: Timeline of the far future. --Guy Macon (talk) 09:21, 31 January 2012 (UTC)

I think this is not correct to have so exact date and time. We can just write: " 64-bit versions of the Unix time stamp will cease to work at year 292,277,026,596" `a5b (talk) 14:12, 31 January 2012 (UTC)

On what basis? We know the exact second when the 64-bit number can no longer hold the date, so why not give the exact time? --Guy Macon (talk) 18:00, 31 January 2012 (UTC)

Ok. What we have: Signed 64-bit time_t has maximum value of: 9,223,372,036,854,775,807 - easy to calculate in good calculator (2^63-1). But how to convert this into year number; to day and month number (Gregorian calendar), to time? How can I know will it be a Sunday or Friday? What is the expression, or are there Reliable sources which already done this conversion? I was able to find source only for year number. `a5b (talk) 18:49, 31 January 2012 (UTC)

Ah. I understand what you are getting at. I agree - the conversion needs to be sourced or the exact second shouldn't be listed. Let me do some searching and see if I can find a reliable source. Excellent catch! --Guy Macon (talk) 19:05, 31 January 2012 (UTC)

This - a verifying of conversion is only the first step. Second step is to ask - is this conversion legal (exact; well-defined) for such huge amount of time. `a5b (talk) 19:25, 31 January 2012 (UTC)

OK, I added another reference that specifies the rollover to the exact second.

It seems clear to me that the result is legal, exact, and well defined. Could you explain why you think it isn't? --Guy Macon (talk) 09:18, 1 February 2012 (UTC)

The arguments about whether or not there will be a Linux box around, or some equivalent, are somewhat beside the point. There are two ways in which the rollover-date of the 64-bit timestamp could occur:

1. humans or transhumans survive for the next 300 billion years, and are still using 64-bit computer RTC hardware, or simulations thereof. Uh oh, WP:CRYSTAL applies, we cannot know that. However...
2. current-day simulations of the future, for instance astrophysics simulations that predict galactic collisions. Not WP:CRYSTAL, these exist now.

But besides the simulating-the-future-today argument, there is the straightforward policy-argument. Reliable Sources often mention the 292-jiggowatt figure, because it is WP:NOTEWORTHY in and of itself, as a huge honking kew1 number. Also, because Some People are anal about datatypes. It's not a personal attack, if it is self-deprecating humour.  :-)   Anyhoo, point being, we wikipedians are *not* predicting the future in violation of WP:CRYSTAL here, by specifying the theoretical rollover-date-in-the-future-leptic-Gregorian-system, because folks can run astrophysical-simulation-software today which will hit that rollover point, as well as various computer-limitation-tutorials that demonstrate the Y2k38 bug and then the Y292jiggowatt bug just for spite.

  At the end of the day, though, all that matters is that we have Reliable Sources out the wazoo which report the 292-figure, that makes the number itself WP:NOTEWORTHY, and therefore wikipedia should list it. Since it's been challenged, give it an inline-cite, plus maybe preface the sentence with "theoretically" if that isn't editorializing. If we have multiple conflicting reports of the exact figure, do what Mariah Carey does, and describe the conflicting birthyear-sources with footnotes. Hope this helps. 74.192.84.101 (talk) 00:57, 5 December 2013 (UTC)