Leap Seconds

I posted a comment in the soft/hard code discussions noting that a minute doesn't have 60 seconds -- you can't hard code that number if you want to be correct. But you can't soft code it either since it's not a constant.

It's an example of programming errors committed by people who don't even know they are programming. Astronomers seem to think that they understand how humans represent time so they go round doing damage with the best of intentions and the least understanding.

I've belabored this point on http://www.frankston.com/public in comments on leap seconds. In terms of this forum it would be interesting and perhaps even useful to look at examples of the damage done by those who don't even know there are concepts they should understand.

Of course it’s unclear where to draw the line considering I view intelligent design as a result of seeing the one example we have and assuming it’s the only possible implementation.

ammoQ

@BobFail said:

I posted a comment in the soft/hard code discussions noting that a minute doesn't have 60 seconds -- you can't hard code that number if you want to be correct. But you can't soft code it either since it's not a constant.

In many cases, the 1 minute = 60 seconds approximation is good enough. For example, when you download a large file over a slow line and the calculations says it will take another 1000 seconds, users will be happy with "time left: 16 minutes 40 seconds" even if a leap second is bound to happen within the next 1000 seconds.

kirchhoff

@ammoQ said:

In every case, the 1 minute = 60 seconds is enough.

There, fixed that for you.

There is no such thing as a non-60-second minute. There is a notational 61-second minute every so many years that is only of interest to implementations of gmtime() in C libraries. This is the only time you will ever see this, and again, it is only a NOTATIONAL CONVENTION.

You will NEVER represent a time remaining or difference in times as having anything other than 60-second minutes, even if the start and end times cross a leap second boundary. If you can't see why this is, then you need to stop writing time/date code and hang up your developer hat.

ammoQ

</sarcasm> g

kirchhoff

@ammoQ said:

</sarcasm> *g*

Oh, that's cool. I get it now that you added the emoticon. Normally when you post all I hear is "let me tell you what should be obvious to a 3rd grader", so you can imagine how I might have been thrown off there.

lol j/k

Random832

@BobFail said:

I posted a comment in the soft/hard code discussions noting that a minute doesn't have 60 seconds -- you can't hard code that number if you want to be correct. But you can't soft code it either since it's not a constant.

There are exactly sixty seconds in a minute. That 00:00:30 may not be exactly one minute later than 23:59:30 of the previous day is immaterial to that. You are confusing time with timekeeping. (incidentally, a month is not a measurement of time at all, and a year as a measurement of time does not contain an integer number of days. A day is 86 400 seconds, a week is 604 800, and a year is 31 556 952 seconds. Exactly.)

In other news, you composed your post in MS Word.

AssimilatedByBorg

@Random832 said:

There are exactly sixty seconds in a minute. That 00:00:30 may not be exactly one minute later than 23:59:30 of the previous day is immaterial to that. You are confusing time with timekeeping. (incidentally, a month is not a measurement of time at all, and a year as a measurement of time does not contain an integer number of days. A day is 86 400 seconds, a week is 604 800, and a year is 31 556 952 seconds. Exactly.)

Your definitions of year and day are highly arguable. A day having 86,400 seconds only works if you define a second as 1/86,400 of a day. And even then, is that a solar day or a siderial day?

If you follow the SI definition of second... well then the number of seconds in a day (sidereal or solar) can vary, and are certainly not exactly 86,400 seconds.

Similar comments apply for "year".

All speaking as astronomers do, of course. Which, most of the time, doesn't relate one bit to what the rest of us Earthlings do.

asuffield

pi seconds is a nanocentury. The rest is just quibbling over the fifth significant figure.

Random832

@AssimilatedByBorg said:

@Random832 said:
There are exactly sixty seconds in a minute. That 00:00:30 may not be exactly one minute later than 23:59:30 of the previous day is immaterial to that. You are confusing time with timekeeping. (incidentally, a month is not a measurement of time at all, and a year as a measurement of time does not contain an integer number of days. A day is 86 400 seconds, a week is 604 800, and a year is 31 556 952 seconds. Exactly.)

Your definitions of year and day are highly arguable. A day having 86,400 seconds only works if you define a second as 1/86,400 of a day. And even then, is that a solar day or a siderial day?

It's a temporal day. which is 794 243 384 928 000 cycles of microwave light absorbed or emitted by the hyperfine transition of cesium-133, and not a fraction more or less. Any connection to the rotation of the earth or to calendar-based timekeeping is historical in nature. The very fact that the other kinds of day are variable, as you said, is why they are not used to measure time.

vr602

If you're interested in Time, the measurement of Time, and the philosophy of the measurement of Time, I recommend this book. It's a novel actually, and a very moving one, but raises and answers a lot of interesting questions about what Time is.

Amazon.co.uk

Also, this raises some interesting Time problems:

http://www.longnow.org/projects/clock/