On Mon, Dec 25, 2023 at 12:24:55AM +0000, Albretch Mueller wrote: > On 12/25/23, David Wright <deb...@lionunicorn.co.uk> wrote: > > On Sun 24 Dec 2023 at 23:05:53 (+0000), Albretch Mueller wrote: > >> On 12/18/23, Max Nikulin <maniku...@gmail.com> wrote: > ... > >> Why would %S be in the range > >> second (00..60), instead of (00..59)?: > > > > Leap seconds—see the example already in the thread: > > https://lists.debian.org/debian-user/2023/12/msg00976.html > > So, a possible (the only?) solution to those kinds of problems would > be to always and explicitly use UTC, right? Or, using the longitude 20 > West (just crossing Iceland, which is 60+ North) or 170 West (too > close to "Vladimir Putin") where so few people live that I don't think > that anyone would care about day time savings or any of that. >
Yes - that's the obvious way. I set my machines to /etc/UTC (or /etc/GMT) and leave them there. No daylight saving time, no offsets - all logs unambiguous. That's why (worldwide) radio logkeeping is/was in UTC. If you're travelling in an aircraft, you don't _need_ to know ground time but you do need to know flight time against a reference time. The Royal Air Force keep to UTC wherever they are in the world for just this reason. Anything else is an offset against the reference: if all your Linux boxes have timestamps against the epoch - which can also be related to a human time if you *have* to - you have a reference there.. The problem comes when someone gives you logs that are taken against a different reference. (See also mapping against Greenwhich meridian (UK) and Paris meridian (France) for many years - two sets of maps that aren't *hugely* different on a world scale but locally very different). > All kinds of software keep time diffs. I am trying to use it in an > obvious human readable way right in the file names. > > lbrtchx >
