Tangential, but what would you need to keep track of time on a different planet, from a computing perspective?
This is the best argument in favor of starting a nuclear war.
I’m not sure I get it, sorry.
If we obliterate human society we don’t have to worry about interplanetary timezones
Hah! I thought you meant to use it as an epoch marker. But you’re right, having to worry about timezones + relativity makes my brain explode just thinking of it.
TAI as a common denominator or lingua franca seems like a decent start.
TAI should be used on Earth too… hopefully libtai will get more traction.
On a different planet is relatively easy. Pick a fixed epoch point and something in the style of UNIX time (elapsed seconds since the Epoch) is fine. The point picked by the UNIX Epoch isn’t magic. OpenStep defined the ‘reference date’ (epoch) as 1 January 2001, GMT and encodes this as a double, which gives microsecond precision over a range of about 285 year before and after the epoch and gradually degrades after that. You can then translate that fairly easily into whatever local calendar you want. We already need to handle calendars with different year lengths on Earth, adding ones with different day lengths isn’t conceptually any harder.
The real difficulty happens if you need to track time on multiple planets, because relativistic effects mean that time will flow slightly differently for people on different planets (general relativity) and travelling between them (special relativity) and so there isn’t an objective measure of time that you can use. If you want to define a sequentially consistent ordering of events that happen on multiple planets that’s going to be difficult (even between here and Mars if you want high enough precision. The relativistic effects from the orbital velocity of GPS satellites is already measurable and that’s a small distance and a fairly low speed).
One wonders how much time synchronization would matter on an Earth-Mars communications network. The time diff is what? 8 light minutes? People would be using store and forward emails for most communication, with voice/video constrained by the time delay and bandwidth.
So it would be enough to know the time on another planet at around 10m intervals or so, rather than the microsecond.
Time synchronisation generally matters for defining serialisation in distributed systems. If your RTT is measured in tens of minutes, then you don’t want to do this with a live two-phase commit, so you almost certainly want some form of eventual consistency. If there is no such thing as a global view of time, this becomes a fun problem.
Wikipedia has an article about timekeeping on Mars, which is probably the most current example: