On Sun, Jan 25, 2009 at 01:58, Azalais Aranxta <[email protected]> wrote: > On Sat, 24 Jan 2009, Philip Newton wrote: > >> On Sat, Jan 24, 2009 at 04:14, Azalais Aranxta <[email protected]> wrote: >> > (We could live with that, if the software would allow us to >> > actually date things in the 1940s >> >> Which reminds me -- is January 2038 considered "the end of time" as on >> LiveJournal? >> >> IIRC, Dreamwidth will be 64-bit throughout, so hopefully their time_t >> (or whatever is used) will also be 64-bit and will not suffer from >> Y2.038K. > > I'm not actually sure that's the reason we can't put > game-chronology timestamps on things, because 1942 is after 1939. > If LJ allowed dates between 1939 and 2038,
Ah, no it's not a 100-year window on 1939-2038. It's based on the fact that many computer systems represent time as seconds since 1 January 1970 00:00:00 UTC (I think), and that many of those computer systems have 32-bit signed time counters, which means that the highest time they can represent is 2^31-1 = 2147483647 seconds since 1970, which is on 19 January 2038. The earliest time, if you allow negative values, is 2^31 seconds before 1970, which is on 13 December 1901. If you don't allow negative values, the earliest time is 1 January 1970. (That's UTC -- so afternoon/evening times on 31 December 1969 would be representable in timezones west of UTC such as the US's time zones.) If you have 64-bit time counters, then the highest date you can represent is about 300 billion years in the future. (And the earliest date about 300 billion years in the past. Whether the "epoch" -- the point in time that you count seconds from -- is 1970 or 1852 or 0 becomes just about irrelevant with such huge timescales.) Cheers, -- Philip Newton <[email protected]> _______________________________________________ dw-discuss mailing list [email protected] http://lists.dwscoalition.org/cgi-bin/mailman/listinfo/dw-discuss
