I have a lot of work, so here is a short post:

Real computers use UnixTime. UnixTime is a 32-bit signed integer representing seconds. 0-Time is 11970-01-01T00:00:00Z HC (HoloceneCalendar). This has 3 problems:

# 0. 1 problem with is is that it only can represent 4,294,967,296 seconds.
# 1. Another problem is that on 12038-01-19T03:14:07Z HC, it runs out (the next second will be 11901-12-13T20:45:52Z HC.
# 2. A 3rd problem is that rather than an invariant count of seconds, seconds repeat for leap seconds.

One can get around the 1st 2 problems by using a 64-bit integer instead with problems with backwards compatibility, but the 3rd problem still exists.

Second timestamps are not good enough for filesystems and databases. One needs timestamps of at least 1 milisecond, with many systems using timestamps of microseconds for futureproofing. Leap seconds play havoc on these timestamps.

I have a proposal solving all of the aforementioned problems:

Use either a signed 256-bit integer or a 243-trit balanced ternary integer represent the number of PlanckTimeUnits since 00000-01-01T00:00:00Z HC (Assume no leap seconds before 11972 HC) in Barycentric Coordinate Time in an invariant count (the computers will use tables in a database for converting this to human-readable time). Use this invariant count of PlanckTimeUnits in timestamps for filesystems and databases.

Please discuss.

I anticipate that somepony will point out that the number of bits require is 4 times larger than 64 bits and that others will object to having to multiply each tick from whichever periodic phenomenon used by an insane number to get the number of PlankTimeUnits to add to the integer. Please let these and other criticisms fly.

I see a problem with using Planck time, namely that we don't know exactly how many Planck times there are in a second, and that our current understanding of physics may change which may show that Planck time isn't as fundamental as we thought. A solution would be to use the current definition of a second; that is, have time be based on period of the radiation corresponding to the transition between the two hyperfine levels of the ground state of the caesium-133 atom. This has an advantage, that it is easy to accurately determine the number of time units that pass in a time period, as we already have atomic clocks and it defines the second.

6136732

Sorry for the late reply, but I work like a DoggyShakeSpeare.

> “I see a problem with using Planck time, namely that we don't know exactly how many Planck times there are in a second,”

We know the number to 3 SigFigs. That is not great, but it is good enough. We can upgrade the definition as we refine it. Since this will be an invariant count, we shall need a database of definition-changes for converting into human-readable time&date.

> “and that our current understanding of physics may change which may show that Planck time isn't as fundamental as we thought. “

Unfortunately, we cannot wait for perfect knowledge.

> “A solution would be to use the current definition of a second; that is, have time be based on period of the radiation corresponding to the transition between the two hyperfine levels of the ground state of the caesium-133 atom.”

What we need are the help of a Chronometrist and a Physicist. I do not have know what is best.