Prev: iTunes - sync Gmail contacts ?
Next: Bookmark Syncing
From: chris on 4 Mar 2010 04:26 On 04/03/10 06:32, Chris Ridd wrote: > On 2010-03-04 00:37:51 +0000, Richard Tobin said: >> If there are still x86 32-bit systems around by then (which wouldn't >> surprise me - the 80386 was introduced 25 years ago, and there's only >> 28 years left), a workaround would be to change the interpretation of >> large negative times, making the range (say) 1950-2087, instead of >> 1901-2038. > > The main problem is that time_t values of -1 are defined to indicate an > error (which is the only reason time_t is signed), so you'd need to > figure out a way to avoid that. > I was wondering why they used a signed int. An unsigned int would have been more sensible, but the error checking makes sense.
From: Richard Tobin on 4 Mar 2010 04:51 In article <hmnubk$fqe$1(a)news.eternal-september.org>, chris <ithinkiam(a)gmail.com> wrote: >I was wondering why they used a signed int. An unsigned int would have >been more sensible, but the error checking makes sense. Why wouldn't you want to represent dates in the past? -- Richard -- Please remember to mention me / in tapes you leave behind.
From: Chris Ridd on 4 Mar 2010 05:12 On 2010-03-04 09:51:29 +0000, Richard Tobin said: > In article <hmnubk$fqe$1(a)news.eternal-september.org>, > chris <ithinkiam(a)gmail.com> wrote: > >> I was wondering why they used a signed int. An unsigned int would have >> been more sensible, but the error checking makes sense. > > Why wouldn't you want to represent dates in the past? You're never going to be able to represent arbitrary dates using a built-in integer type. Use struct tm instead. -- Chris
From: R on 4 Mar 2010 05:24 Peter Ceresole <peter(a)cara.demon.co.uk> wrote: > Well, checking Palm desktop running (very nicely) under OS10.4.11, in > mid-Feb 2040 it wraps round to 1904. That's true of real time, too. Physicists haven't noticed.
From: R on 4 Mar 2010 05:29
Richard Tobin <richard(a)cogsci.ed.ac.uk> wrote: > Unix uses the number of seconds since Jan 1 1970. It traditionally > stores this in a 32-bit signed integer, which will overflow in > 2038. Presumably this will have changed to a 64-bit integer long before > then. I'm running the 64-bit kernel. Would that use 64-bit time? |