Re: localtime()
Laurent Bercot <[email protected]> Mon, 5 Dec 2011 00:45:51 +0100
| Newsgroups | gmane.linux.lib.dietlibc |
|---|---|
| Message-ID | <[email protected]> |
> Nope, I'm only an amateur who happens to enjoy making a few small programs for > home use that are immune to software upgrades that can go wrong, I'm not in the > same league as you guys. Anyway, I was not sure what is the officially expected > behavior, the linux man page is not that clear. Subtracting 24s (each second, > in a "wall-clock" program) is not ideal but it does the job. Correct time handling under Unix is a mess. A comprehensive guide to understand the problem and its solutions can be found here: http://www.madore.org/~david/computers/unix-leap-seconds.html POSIX more or less requires people to set their system clocks to UTC: so a Berliner complying with POSIX would set his clock to UTC and his timezone to Europe/Berlin to get the correct local time. But this setting breaks in several subtle ways, so a few people, including DJB, Arthur David Olson (author of the right/ timezones) and me, recommend disagreeing with POSIX and setting system clocks to TAI-10 and using right/ timezones to get the local time - so a Berliner setting his system clock to TAI-10 should have a timezone containing right/Europe/Berlin. What is this thing about German law ? Does German law require a specific system clock setting ? If it requires using the right/ timezones, that means the system clock should be set to TAI-10, not UTC. -- Laurent