Samoa and Tokelau Are Skipping December 30th

ocean_soul writes "Starting January 1, 2012 Samoa and Tokelau will be in time zone +13 instead of -11. This means there will be no December 30, 2011 in these countries. The decision to switch time zone was based on the changing international business relations of Samoa. Samoa had adopted the -11 time zone to make business with the U.S. easier. However, currently Samoa's most important trading partners are Australia and New Zealand. By switching time zone the work-weeks and week-ends on Samoa and Tokelau will be synchronized with those in Australia and New Zealand."

Comments

SIgn of the "times"

[LostCluster]

DST changes in the USA caused times to be odd on certain devices such as VCRs to incorrectly make the change... what's Samoa's tech devices thinking for time zone updates or will everybody have to do a lot of twisting to their watch. For anybody with any interest in what goes on there this is a big tech story.

Re:SIgn of the "times"

[sconeu]

No change in time. -11 = +13 mod 24.
The only change is the date.

Re:SIgn of the "times"

[InterestingFella]

But how does Thailand handle unixtime? It's already year 2555 there. That's way past 32-bit int.

Re:SIgn of the "times"

[tepples]

But how does Thailand handle unixtime?

UNIX time is seconds since 1970-01-01 00:00 UTC minus the accumulated count of leap seconds. Thailand just counts it using days since whatever Thai date corresponds to the Gregorian date 1970-01-01.

It's already year 2555 there.

Yeah, and it's 5772 in Hebrew years, which roughly correspond to the time since Adam and Eve were created.

Re:SIgn of the "times"

[LostCluster]

This makes me wonder. Are people going to be paid/charged interest for a non-existing 12-30-11 there?

Re:SIgn of the "times"

[InterestingFella]

If you go to north pole and keep running around the pole in same direction, crossing timezones, you can go infinitely back or forward in time!

Re:SIgn of the "times"

[Cyberax]

Sure it does. A lot of countries change DST rules all the time.

Re:SIgn of the "times"

[ceoyoyo]

Yes. Apparently they're paying people who were scheduled to work on Dec. 30. I assume they'll charge interest too.

Re:SIgn of the "times"

[ceoyoyo]

Until you hit that darn international date line.

Re:SIgn of the "times"

[mj1856]

Does current time code even have the sufficient smarts currently to handle specific countries CHANGING their TZ on a particular date?

Yes. Linux/Unix has a long history of tracking timezone changes for specific countries, states, provinces, etc. It's called the Olsen Timzone Database. It was recently taken over by IANA, and is hosted here http://www.iana.org/time-zones

They are discussing this specific issue here:
http://mm.icann.org/pipermail/tz/2011-December/008458.html

This makes me wonder. Are people going to be paid/charged interest for a non-existing 12-30-11 there?

It depends. I work for a time and attendance company as software developer, so I have some insight. Basically, this is handled just like a DST change, but for a much longer period.

Many timekeeping systems (hardware and software alike) just keep track of "local time". Some have the ability to keep a list of DST changes that need to be applied at specific times, and some use NTP or other protocols to sync their clocks and pickup timezone changes that way. While these systems handle "spring-forward" changes ok, they are usually flawed in the way they handle "fall-back". If someone clocks in or out DURING the fall-back period, there is no way to tell if they get an extra hour or not, because there is no recorded distinction between the two times that are both called the same thing. The good thing about DST is that the change usually happens in the middle of the night, which minimizes the number of manual corrections that have to be made.

The solution to all of this, of course, is recording time as UTC and converting it for proper display depending on context. Some systems out there caught on early, but really this idea is just now making its way into the market. This is where the timezone database is very valuable. Windows also has a timezone database (different than the Olsen DB), but Microsoft only pushes it out every few months (via windows update), so it is often behind in various parts of the world. Microsoft timezone info here: http://blogs.technet.com/b/dst2007

Since Samoa and Tokelau are skipping a day, this is a "spring-forward" scenario - which is very easy to calculate. It is highly unlikely that they will have issues with paying an extra day (or charging an extra day's interest), as long as they consider the change like any other DST change. I would think that this is big news there, so anyone with custom code will probably be aware of the situation and make the correction.

Of course, if you have a bank account in another country, they are going to say a big "screw you" to your request to be charged one day's less interest just because your homeland is skipping a day. :)

Re:SIgn of the "times"

[Anonymous Coward]

Just jump over it.

Re:SIgn of the "times"

[mj1856]

What do commonly used OSs, apps, do about a user that is entering NONEXISTENT dates, like Dec 30, 2011 *in that locality* (but not necessarily in the locality of the user)? I guess dates are meaningless unless accompanied by the locality of the date.

Exactly. There are a whole slew of dates/times that simply do not exist in certain timezones. As an example, March 13, 2011 at 2:30 AM in Eastern Time (usa), which is smack in the middle of a "spring-forward" DST change. Say you use the TimeZoneInfo class in .Net Framework to convert from this non-existant time to UTC:

(c# code)
        var tz = TimeZoneInfo.FindSystemTimeZoneById("Eastern Standard Time");
        var localTime = new DateTime(2011, 3, 13, 2, 30, 0);
        var utcTime = TimeZoneInfo.ConvertTimeToUtc(localTime, tz);

You will get an ArgumentException with the message "The supplied DateTime represents an invalid time. For example, when the clock is adjusted forward, any time in the period that is skipped is invalid."

I'm assuming other languages/frameworks/os's do similar things, either throwing an exception, or returning null, NaN, or similar.

In regards to regular updates, Windows gets timezone updates from Microsoft via Windows Update. According to Wikipedia, Linux updates of the TZ database are different depending on distribution: http://en.wikipedia.org/wiki/Tz_database#Maintenance

Re:SIgn of the "times"

[mjwx]

But how does Thailand handle unixtime? It's already year 2555 there. That's way past 32-bit int.

Considering that Unixtime only started in 2514.

Epoch fail.

Your math does not calculate

[InterestingFella]

If they change on January 1, then December 30 would already been passed at that point. So, that would mean it's already Dec 30 there right now and it cannot be Jan 1 yet according to any time zone. The math is fail.

Re:Your math does not calculate

[Yvan256]

I'm more confused about the fact that my calendar tells me there's a december 31st. Are they going to go "december 29th, december 31st, january 1st"?

Re:Your math does not calculate

[owlstead]

"This means there will be no December 30, 2011 in these countries." You've just shown that you can skip calendar dates, congratulations!

Re:Your math does not calculate

[InterestingFella]

Bah, you westerners with your stupid one new year. I will celebrate western new year tomorrow, then in January I'll be in China when it's Chinese new year, and finally in April I will be in Thailand when it's Thai new year. Yay, three new year parties for me!

(and the Songkran will have country-wide free-for-all water fight, with insane amount of hot girls and ladyboys in wet t-shirts. Woohoo!)

Re:Your math does not calculate

[Anonymous Coward]

We showed that we can skip calendar dates, on several occasions. E.g. when the 'catholic' world switched from the Julian to the Gregorian calendar in 1582-4. Then most of western Europe and the Americas in 1752-3. Eastern Europe in 1918, etc.

http://www.searchforancestors.com/utility/gregorian.html

Re:Your math does not calculate

[Penguin]

.. and cal even supports it:

$ cal 9 1752
September 1752
Su Mo Tu We Th Fr Sa
1 2 14 15 16
17 18 19 20 21 22 23
24 25 26 27 28 29 30

... however it's not that effective as locale is not taken into consideration. As your link mentions, "only" England+Scotland+colonies switched at that point.

Re:Your math does not calculate

[catbutt]

Probably because dec 31, 2011 is in lots of legal contracts. If you eliminate that date, you have a lot more chances for people to try to weasel their way out of contracts, and more work for courts.

This happened to me once

[Stargoat]

This happened to me once. I crossed the International Date line on December 24. It was December 26 on the other side. It was the year without a Christmas.

Re:This happened to me once

[Anonymous Coward]

Go the other direction. I was on a ship where we were scheduled to have thanksgiving day twice. Seeing how the food was rather mediocre, we naturally assumed this meant we would get two thanksgiving dinners, which would have been wonderful. The ship then rescheduled the repeat day earlier (when you have 10 days between the two ports when you're making the idl crossing, you can schedule it whenever you want).

Re:This happened to me once

[archen]

Marry someone from Canada and you can have Thanksgiving twice a year every year (or the other way if you are Canadian).

Re:This happened to me once

[nedlohs]

I skipped over my wedding aniversary on my flight to Australia last year. Best anniversary so far!

Given wanting to do the flying on the weekend I won't be able to repeat that until 2016.

Re:But...

[eclectus]

Yeah, they should have skipped a Monday. Or even a Wednesday

Re:fail

[TamCaP]

just count on your knuckles from left to right (both hands); each big knuckle - {31}, each small knuckle - [28-30]

First Break in the Seven Day Week Cycle

[dbkluck]

According to wikipedia (admittedly with a "citation needed") the seven day week cycle has continued unbroken for almost two millenia, despite numerous readjustments in the date over the centuries. So although skipping even a whole bunch of dates is not unheard of (e.g., Thursday, October 4th, 1582 followed immediately by Friday, October 15th when the Gregorian calendar was adopted), this seems like the first time in a long time that the day after Thursday hasn't been Friday.

Re:First Break in the Seven Day Week Cycle

[__aajfby9338]

Once I injured my back, and my doctor prescribed a collection of meds that caused Monday to be the day after Friday.

Try "cal 9 1752" at a *nix command prompt

[knorthern knight]

Good thing they didn't have VCRs back then.

Re:fail

[TamCaP]

s/small knuckle/valley between knuckles/ :-)

New Zealand is important!

[SplashMyBandit]

.. for once.

[Disclaimer: I live there]

Re:New Zealand is important!

[roc97007]

For once? The geekiest movie trilogy of all time was filmed there.

Not really, no.

[Anonymous Coward]

No, actually it's not.

First, UT is a bit confusing. You have to specify which UT you mean: UTC, UT0, UT1, UT1R, etc...

All these, except UTC, are based on celestial movement. Which means they will vary due to natural causes. A slight wobble in Earths orbit or a little bit of tectonic shift will cause seconds to be shorter or longer. UTC is based on 'artificial' timing (atomic timekeeping) and as such has slightly different seconds than the other UT's.

So, no, UTC is not simply another UT without counting leap seconds.

Timekeeping is hard. Really hard.

Re:Changing trade relationship

[NJRoadfan]

That would be American Samoa, which looks to be retaining the UTC-11 time zone.

No the key change is the work-week.

[kale77in]

Their work-week is now in sync with Aust and NZ, rather than only having four days that coincide.

Re:No the key change is the work-week.

[SteveTheNewbie]

My goodness, three days ? let me draw you a little table.

M - S
T - M *
W - T *
T - W *
F - T *
S - F
S - S

The days with a * next to them are 'working days which coincide', you'll notice there are four of them. I'm quite amazed that you got modded to 4.

Why not wait for leap day?

[AthanasiusKircher]

If they just waited two months until February, they could have just issued calendars that skipped leap day, and few people would have even noticed (until the work week suddenly became shorter).

And they'd get to say that they were "leaping" over leap day....

Re:But...

[Megane]

If the wanted to skip a Monday, they could have just shot the whole day down.

Re:Kill timezones already

[hankwang]

Switch the entire world to UTC (and kill am/pm since they'll no longer correspond to morning/night in half the world). ... just the idea of eliminating "2pm your time or mine?" makes it worthwhile.

That might be convenient for making appointments for telephone conferences, but it really sucks if you actually travel to such a timezone and need to schedule your daily program; then you will have to calculate the offset relative to your old place every time you wonder whether it is already lunch time, or whether the shops/offices are open. Not to mention that having the date and day of the week change in the middle of the day might also be rather inconvenient: what does "see you on Wednesday" mean?

And as for appointments: calendar applications already take care of calculating the time zones while scheduling meetings.

Inquiring minds want to know

[jabberw0k]

What will this do to the supply of Girl Scout Samoa cookies? (For the record, I hoarded Manila folders when the Marcos government fell.)

Let me correct this

[tepples]

I apologize for sidetracking the discussion to the leap second issue. Let me try again: "seconds since the UNIX time epoch, counting any leap seconds as 0 instead of 1". This makes each midnight-to-midnight period 86400 seconds long. My point is that the UNIX time epoch is the same regardless of the local civil calendar's epoch.

Re:NIGHTMARE

[Anne Thwacks]

As a software developer i assure you, every day is a nightmare coming true.

Thats fixed it for you!

Re:Fine idea

[lgw]

Presumably this will be decided by how much alcohol you consume on 12/30, and when you regain consciousness?