News for hackers. Stuff that matters.
Re:Ask HN: What is the oldest, still supported OS? by WorldMaker on Monday February 28, 2022 @05:03PM (#30505780)
It's often used to encourage long-term thinking. I most directly associate it with the Long Now Foundation [1], which encourages that format for dates as many of their initiatives strive to consider the Y10K view and beyond.

[1] https://longnow.org/

Re:Ask HN: What is the oldest, still supported OS? by imoverclocked on Monday February 28, 2022 @05:30PM (#30506059)
Since we are adding insignificant digits, I think there aren't enough zeros after the most significant digits; We need to focus on solving todays problems too.

For an extreme version of this, 1961 might be written in scientific notation: 0.196100e4

This is also future-proof as we will always have a built-in 0 at the beginning of the number implying that we never get to "1" and there will always be more time to solve the problems we can't even begin to understand that may exist 10k years from now.

Sorry for the sarcasm, it was the easiest way to make my point.

Re:Ask HN: What is the oldest, still supported OS? by eesmith on Tuesday March 01, 2022 @02:43AM (#30510897)
RFC 2550 at https://datatracker.ietf.org/doc/html/rfc2550 points out "Floating Point Is No Solution" in §1.2.3.2, and earlier stresses the importance of lexical sorting. The authors propose a time encoding scheme valid up to the year 10 ** 18308.

(Hint: look at the publication date.)

Re:Ask HN: What is the oldest, still supported OS? by kragen on Wednesday March 02, 2022 @12:41AM (#30524164)
Oh dear, I never thought about the year 922,337,203,685,477 problem before.
Re:Ask HN: What is the oldest, still supported OS? by eesmith on Wednesday March 02, 2022 @05:27AM (#30525811)
It's important if we want our time system to survive until the Degenerate Era. https://en.wikipedia.org/wiki/Future_of_an_expanding_univers...

128 bits isn't even enough to get to the putative half-life of proton decay.

There's some truly mind-boggling numbers in https://en.wikipedia.org/wiki/Future_of_an_expanding_univers... "the non-zero probability of producing a new Big Bang in roughly 10^10^10^56 years"(!).

These numbers are far beyond https://datatracker.ietf.org/doc/html/rfc2550#section-2.4.1 , which is only 10^14; about the time star formation ceases.

Time for an updated RFC!

Re:Ask HN: What is the oldest, still supported OS? by WorldMaker on Monday February 28, 2022 @09:21PM (#30508469)
Really, it's an epoch problem of sorts. If we picked an epoch date with a longer view we might have a better feeling for it as large numbers. For instance we are roughly somewhere between Y12K and Y13K since the estimated start of the Holocene. On the other side, it might feel more scientific to pick a far future epoch date and negative numbers would be a greater reminder of time running out, such this being roughly Y-5e9 until the expected death of our sun.

(To be fair to the Long Now, not that they necessarily need fairness, and I'm not a member I just read about them here on HN mostly, adding the initial insignificant zero was intentionally an easy "hack" to keep things recognizable and not get into quagmires like changing the epoch date.)

Re:Ask HN: What is the oldest, still supported OS? by ianburrell on Monday February 28, 2022 @09:46PM (#30508712)
There is the Holocene calendar which adds 10,000 to the year. The current year would be 12022 HE. It has advantages that gets rid of BC/year zero confusion and that all historical dates are positive numbers.
Re:Ask HN: What is the oldest, still supported OS? by kragen on Friday March 04, 2022 @09:34PM (#30563555)
To me this seems like using the Unix time convention but changing the epoch time from midnight UTC to midnight at UTC-12:00, the timezone of Baker Island: it's close enough to produce confusing ambiguity (is "2020" or "02020" intended to mean two years ago, or an estimated date when Jericho's pre-pottery Neolithic A wall fell out of use, suggesting an abandonment of the site?), and it makes it more likely that your software will have untested cases with bugs in them.

In the case of the Holocene Calendar the bugs would just be a question of negative dates; in the case of UTC-12:00 it would be a nonnegative epoch time that is nevertheless in 01969 instead of 01970, which will only happen if someone adopts a more westerly timezone like UTC-13:00, pretty much guaranteeing you won't test that case beforehand. (Samoa did something similar in 02011, but it was in the other direction.)

Of course, as other people have pointed out, you shouldn't use fixed-width date fields in your software.

Re:Ask HN: What is the oldest, still supported OS? by runlevel1 on Monday February 28, 2022 @09:24PM (#30508496)
If we're going decimal I'd prefer it be the percent of time elapsed between the Big Bang and heat death of the universe.
Re:Ask HN: What is the oldest, still supported OS? by imoverclocked on Monday February 28, 2022 @11:35PM (#30509630)
The problem is understanding and measuring either of those events with any precision. Even if we somehow could, time also seems to move faster/slower depending on the strength of a gravity field in which it is being measured. I think this makes it really hard to pin down dates near super massive objects... and also means that our linear perspective of time completely breaks down on this scale. eg: where is the reference time located for a few billion years just after the big-bang?
Re:Ask HN: What is the oldest, still supported OS? by kragen on Monday February 28, 2022 @05:36PM (#30506129)
This is a great idea! I'll try it next week if I remember.
Re:Ask HN: What is the oldest, still supported OS? by mattkrause on Monday February 28, 2022 @05:49PM (#30506253)
Next 0.1923e-1, you mean?
Re:Ask HN: What is the oldest, still supported OS? by kragen on Monday February 28, 2022 @07:20PM (#30507248)
Next 0.1917e-1!