Hacker News Reader

Comment by jylam | Hacker News Reader

It seems I've got an unpopular opinion reading the comments here, but having some leap seconds from time to time ensure we are able to manage them. But "no leap second should be added for at least a century" ensures there will be a y2k reckoning every century. Seems short sighted to me, even if it is not a game-changing issue, let's be honest.

View on HN

Why have them at all? how many people do things where they actually matter? at the current rate it'll be ~9000ad before we hit an hour offset, and we do hour offsets twice a year, so if an hour off is okay why is 1 second off not?

Ok let future generations take the hit. Got ya. Selfish ...

How big of a hit is it on the future generation, they can just change their DST, which already happens every few years anyways?

The critical question is whether they'll have to leap forward or fall backward. We simply cannot abide stealing an hour of sleep from our cyborg/transhuman descendants. We'd never live it down.

When longtermism meets leap seconds.

Hopefully by the time they need to make an hour's adjustment, we'll have done away with the anachronism that is DST. So then they'll have to learn to do it all over again from scratch.

> at the current rate it'll be ~9000ad before we hit an hour offset

And the leap year problems of the Julian calendar weren't a problem… until they were. And then good luck coördinating things:

* https://en.wikipedia.org/wiki/Gregorian_calendar#Adoption_by...

* https://en.wikipedia.org/wiki/Adoption_of_the_Gregorian_cale...

The Julian Calendar was good enough for centuries after the empire that invented it had fallen, which I think counts as good enough in general. That said, I'd love to read a story about tracking down some time bug a million layers of abstraction down from the state of tech in 9000AD.

The thing is, we've remade calendars many times in the past millennium and our current calendar is not gospel. Sidereal drift can and should be addressed with a calendar fix, not a clock fix. The clock's a clock, the calendar's a calendar. It doesn't matter if we can make our clocks calendar-like because they shouldn't be anyway.

> we do hour offsets twice a year, so if an hour off is okay why is 1 second off not?

The actual Unix time does not change when the clocks are switched between daylight saving time and standard time. Only the time zone changes.

When a leap second occurs, the actual Unix time changes, which can lead to bugs, e.g. when a positive time difference comes back as negative. To prevent such issues, a monotonic clock can be used to measure time intervals.

Edit: OK, just re-read some docs, looks like POSIX chose the worst of all options and decrements the timestamp experiencing the same second twice...

Wrong info from before edit: Unix time doesn't change either, it's the number of seconds since 00:00:00 UTC on 1 January 1970, though systems may or may not ignore that and set it to match the UTC time.

Yes. It does. Op is correct. A leap second specifically is an adjustment because “how many seconds since 1970” changes because the earth’s rotation isn’t constant speed. UTC tone definitely changes. If it didn’t then you wouldn’t hear anything about it and it would just be transparently folded into your time zone database.

And that timescale is called TAI. It's always been an option, but for whatever reason, people (and standards) (collectively) use UTC but want it to act like TAI.

By that time we should be an interplanetary species and have a bigger time coordination fish to fry.

The difference should be handled at the same layer as time zones, which is already a mess and is already well-suited to deal with this type of thing.

I'd argue there should be no adjustment until it gets to be +- 15 minutes from UT1. And even then, NO clock skewing, just a time zone adjustment across the board.

Its not a leap once a century thing, it's a rejigger the timezones by an hour every four to five thousand years. (either define new zones one hour over or redefine existing ones)

We already rejigger the time zones from time to time, so thats already handled. It's still a source of issues, sure, but it's once source instead of two.

The timezone rejiggering could also be set hundreds of years in advance... rather than the very short notice we get for leap seconds.

Leap seconds are not useful. We do not need to use them anymore. While yes you are right in that more leap seconds makes it easier to handle leap seconds, it is just better to not have to worry about leap seconds altogother. It is unneeded complexity that helps no one.

Reply on news.ycombinator.com