I disagree, it’s not equal but different. The French famously tried to switch to a decimal system for angles before, but failed in no small part because of the relatively few unique prime factors. Being able to divide evenly by three turns out to be more important than five.
To quote yet another time format: NTP 64-bit timestamp format (rfc8877), which is 32 bits seconds since epoch + 32 bits fixed point second fractions. (Outside of Network Time Protocol, you'll find this baby for instance in ISOBMFF ProducerReferenceTimeBox(prft)).
Here seconds are just 1/(24*60*60) of a day as expected, but the base 2 fixed point part, where a tick "is roughly equal to 233 picoseconds" makes you want to pull your hairs out if you just want to accurately express milliseconds. (Similarly for other timescales frequently used in media processing, like 90kHz, 25, 60 or 29.97)
The answer to all this is of course: hand waving — "you don't need that". Your time can be perfectly accurate in itself (ie. an accurate discrete sample of continuous time), even if no accurate conversion exists to some other time system.
toxik|2 years ago
pipo234|2 years ago
Here seconds are just 1/(24*60*60) of a day as expected, but the base 2 fixed point part, where a tick "is roughly equal to 233 picoseconds" makes you want to pull your hairs out if you just want to accurately express milliseconds. (Similarly for other timescales frequently used in media processing, like 90kHz, 25, 60 or 29.97)
The answer to all this is of course: hand waving — "you don't need that". Your time can be perfectly accurate in itself (ie. an accurate discrete sample of continuous time), even if no accurate conversion exists to some other time system.
pif|2 years ago
gorlilla|2 years ago