DailyDirt: Time, Time, Time. See What's Become Of..
from the urls-we-dig-up dept
With the Apple Watch available now, maybe more people will be interested in wearing fancy watches again — instead of just relying on their phones. Fancy watches once focused on telling time with extreme accuracy, but then digital watches made it really cheap to keep accurate time that was more than “good enough” for most folks. It used to be annoying to need to adjust clocks for daylight savings and power outages, but as more and more clocks are connected to the internet (except for ovens and some cheap alarm clocks), we barely need to think about how to change the time on a clock (who owns a VCR anymore?). Check out these links on accurate time keeping.
- There’s a new atomic clock that’s three times as accurate as the last most accurate clock — losing less than 1 second in 15 billion years. The universe hasn’t even been around for 15 billion years, and this clock probably won’t last anywhere near that long either. Still, it’s nice to have a clock that can measure time dilation effects and other weird phenomena. [url]
- John Harrison designed a clock in the 1700s that was accurate to a second over 100 days — using a pendulum mechanism. Harrison wasn’t able to build a working version of this clock, (and people thought he was crazy when he claimed he could) but modern clockmakers have constructed one from his designs — and it works. [url]
- The 26th leap second is scheduled to be added to some clocks on June 30th — at 11:59:60pm. Try not to rely on any computers that might not be able to handle this leap second, if you can. Or join the movement trying to abolish the leap second entirely, but that probably won’t happen until 2018. [url]
After you’ve finished checking out those links, take a look at our Daily Deals for cool gadgets and other awesome stuff.
Filed Under: accuracy, atomic clock, clocks, john harrison, leap second, pendulum clock, time, watches
Comments on “DailyDirt: Time, Time, Time. See What's Become Of..”
Leap seconds
Writing software to deal with this stuff is not simple. Most developers simply gloss it over, and when everything crashes they simply shutdown and restart all their systems. FWIW, I had to write software that could deal with leap seconds back about 20 years ago, so I know something of what I speak here.
@midnight
Oh no! Their show will have to run extra long on June 30 as it always starts at 11:59 and 59 seconds! Actually, Comedy Central will probably just run an extra commercial.
Using The Extra Second
What are you going to do with the extra second?
I’m going to turn my foreplay into fiveplay!
Leap Seconds are Essential
Leap seconds are needed for the same reason leap days are needed every fourth year: to keep our clocks and watches synced with the Earth’s rotation and its motion around the Sun.
By the way, if you think Linux (& UNIX) systems have a problem with leap seconds that will be a minor hiccup compared to the train wreck looming for January 19, 2038 at 3:14:07 GMT.
Perhaps those looking to abolish leap seconds should get their priorities in order. (Reportedly some systems are going to start showing issues re the 2038 problem as early as 2018.)
Great
Great post! thanks
If the MBAs don’t like the present time keeping system, including leap seconds, then why don’t they just make their own time keeping system and leave the present one as is?
There are benefits to time keeping which is in sych with the star systems, take that away and people will just have to create another one.
Re: Re:
Such a system already exists: TAI. It’s 35 seconds ahead of UTC. Computers should use that instead, and maybe convert to UTC for display.
Some of the people wanting to abolish the leap second want to have a leap hour instead. It’s so ridiculous it’s probably just a trick–hope people forget about it before it’s ever needed (in a few thousand years).
Re: Re: Re:
Never heard of it. Nor has wikipedia. If “TAI” is an abbreviation what does it stand for?
Re: Re: Re: Re:
http://en.wikipedia.org/wiki/International_Atomic_Time
Re: Re: Re:2 Re:
FWIW, it’s an abbreviation in much the same way as UTC is an abbreviation for Coordinated Universal Time. (ie- it’s not)
Can someone explain in plain English, the problem with leap seconds and why they will cause computers to crash?
Maybe I’m just dumb, but I don’t understand the problem. The computer itself doesn’t know about the leap second. Disconnect it from the net and it will happily go on keeping time as if nothing had happened, except that it will be one second ahead of everyone else. Connect it back to the net, it will see that the time is off and reset itself to match everyone else.
I can set my computer to an hour ahead and nothing bad happens other than the timestamps on files being wrong. Nothing crashes.
I read one explanation which claimed that leap seconds were a problem because all computer time is calculated by counting the seconds from something like 1972. What kind of a nutcase would calculate time based on decades worth of seconds? That would be like calculating all monetary transactions in pennies.
Is my digital clock going to explode because the world has added an extra second to the time? Putting aside the fact that my clock doesn’t match up to the world time anyway, all I would need to do is set it back one second and it’s back in sync.
It’s not like they’re adding an extra second to every day. They’re adding a single second at one specific time. Once that second has been added and computer clocks adjusted, time keeping goes back to normal.
OK, calculating a time difference between dates before and after the leap second would carry a single second error, but does that really matter? How many applications in the world absolutely depend on computers being able to calculate the difference between two dates down to the second? And if it matters, the software can be patched to insert an extra second when performing the calculations.
Re: Re:
See eg http://www.theregister.co.uk/2012/07/02/leap_second_crashes_airlines/
It’s not specifically a problem with leap seconds, it’s just that some versions of some software didn’t cope well with NTP adjustments as large as a second, because apparently someone thought it was a good idea to use a non-monotonic clock to implement kernel timers. See http://serverfault.com/questions/403732/anyone-else-experiencing-high-rates-of-linux-server-crashes-during-a-leap-second/403767#403767 for good coverage of the details.
Re: Re: Re:
Thanks. The linked article didn’t make it clear that this was due to a bug. It made it sound as if it was just a shortcoming in the software that couldn’t deal with the time being adjusted.
Re: Re:
I read one explanation which claimed that leap seconds were a problem because all computer time is calculated by counting the seconds from something like 1972. What kind of a nutcase would calculate time based on decades worth of seconds? That would be like calculating all monetary transactions in pennies.
That’s a very real problem that has nothing to do with leap seconds. See http://en.wikipedia.org/wiki/Year_2038_problem
Note that the problem is well underway to being solved in general purpose computers, and mostly remains for specific classes of embedded systems… so is mostly an issue for things that have already been created that will last for another 20 years or more (and care about the date).
Re: Re: Re:
I still think it was a mistake to store the time as the amount of seconds that have passed since some arbitrary date.
Sure, storing the date as a 32-bit number might have been the most efficient way to store the date and time, saving a few bytes, but then they would need to add extra code to convert that 32-bit number back into a human-readable format. If they’d used separate bytes for the seconds, minutes, hours, days, months and year, they could have made up the difference in much simpler code used to convert those bytes into human-readable date formats. At the very least, they should have used a single byte for the year and then they could have stored the rest of the time/date as an offset from that year. That would have given them more than 256 years for the cost of a single extra byte.
“Can someone explain in plain English, the problem with leap seconds and why they will cause computers to crash? “
Poorly written code and an unwillingness to correct it.