20

AFAIK java stores dates in long variables as milliseconds. Consequently someday there will be no value (cause long has a maximum) which will correspond to the time of that instant. Do you know when it will happen?

Denys S.
  • 6,358
  • 12
  • 43
  • 65
  • 8
    I believe the sun is expected to go nova before then. Sun has already gone nova. – Paul Tomblin Nov 16 '10 at 14:23
  • 3
    Actually, I have that wrong - Java Date has 292 million years, but according to http://www.space.com/scienceastronomy/solarsystem/death_of_earth_000224.html the earth has 500 million years. – Paul Tomblin Nov 16 '10 at 14:30
  • Related: http://stackoverflow.com/questions/2978452/when-will-system-currenttimemillis-overflow – BalusC Nov 16 '10 at 23:35

2 Answers2

61

It's easy enough to find out:

public class Test {
    public static void main(String[] args) {
        System.out.println(new java.util.Date(Long.MAX_VALUE));
    }
}

Gives output (on my box):

Sun Aug 17 07:12:55 GMT 292278994

You may need to subtract a bit from Long.MAX_VALUE to cope with your time zone overflowing the range of long, but it will give a reasonable ballpark :)

Jon Skeet
  • 1,421,763
  • 867
  • 9,128
  • 9,194
28

According to the current leap-year regulations the average number of days per year will be

         365 + 1/4 − 1/100 + 1/400 = 365.2425 days per year

This means that we, in average, have 31556952000 milliseconds per year.

The long-value represents the number of milliseconds since the Epoch (1st of January, 1970) and the maximum number represented by a Java long is 263 − 1, so the following calculation

         1970 + (263 − 1) / 31556952000

reveals that this representation will overflow year 292278994.


This can, as Jon Skeet points out, be confirmed by

-> System.out.println(new Date(Long.MAX_VALUE));
Sun Aug 17 08:12:55 CET 292278994
Community
  • 1
  • 1
aioobe
  • 413,195
  • 112
  • 811
  • 826
  • 2
    With such large numbers being used as dates, leap seconds and tidal acceleration probably need to be factored in too. – Qwerky Nov 16 '10 at 14:35
  • 2
    @Qwerky, yeah, I thought about that too, and search for astronomical measurements of the number of milliseconds of a year, but then it struck me that the question was about the java date-implementation, and this does most likely *not* take tidal acceleration etc into account :-) – aioobe Nov 16 '10 at 14:37