Using "midnight" to mean the beginning of the day could be confusing

Fri Jan 22 06:33:32 CST 2016

Because of leap seconds or the precision needed for the last minute our
second of the day, it is admitted to specidy a valid time "24:00" for
specifying the end of a day. Most legal texts however avoid specifying
ending dates, they specfiy the start date (and optional time, 00:00 is then
implicit) after which something has changed.
For monthly or weekly billing periods, it is customary to specify the
ending date of the period, no time is given, the last day is of the month
or week taken inclusively (up to 24:00).
Using "23:59" or "23:59:59" or "23:59:60" should be avoided when you mean
the end of the day

(note that "23:59:60" is perfectly valid but NOT equivalent to "24:00" at
end of every month with a extra leap second, so "23:59:60.00" is still one
second before the end of that day which occurs just before
"23:59:61.000"="24:00";
those leap seconds may happen at end of June or December on some years).

2016-01-22 9:12 GMT+01:00 Jon Skeet <skeet at pobox.com>:

> Just as one extra wrinkle, if we define "midnight" to be "the time that
> one day becomes another", it doesn't always happen at 00:00. If we define
> it to be 00:00, it doesn't always occur.
> I'm basically thinking about time zones (e.g. Brazil) which "spring
> forward" at what would have been midnight, e.g.
>
> 23:59:58
> 23:59:59
> 01:00:00
> 01:00:01
>
> Depending on your definition of midnight, it either doesn't happen or it
> happens at 01:00. (Fortunately I don't *think* there are any examples
> where 00:00 occurs twice due to fall-back rules, although I wouldn't be
> surprised.)
>
> I'm not saying this is something we should fall out specifically - but
> it's something we should be aware of when considering using the term
> "midnight".
>
> Jon
>
>
> On 22 January 2016 at 07:57, Mark Davis ☕️ <mark at macchiato.com> wrote:
>
>> I think it really depends on context. I think the following, for example,
>> refer to the same time, the instant between Tuesday and Wednesday.
>>
>>    - Wednesday, I was wide awake from midnight to 5am.
>>    - Tuesday, the party lasted from 7pm to midnight.
>>
>> The context of a range makes it clear what was meant.
>>
>> I'm ok with holding back on using midnight, except when
>>
>>    1. there is a word for midnight in the locale that (predominantly)
>>    means the start of the day (00:00).
>>    2. in time intervals (where the context is then clear enough).
>>    However, see below.
>>
>> Normally, date-time software views time-periods as half-open intervals.
>> For example, the first hour of a day is from 00:00 to 00:59:59.9..., a day
>> is from 00:00 to 23:59:59.9999..., a year is until Dec 31,
>> 23:59:59.9999..., and so on.
>>
>> #2 is connected with a separate ticket which is to allow for the
>> time-of-day to be *formatted* as being 24:00 or after. The primary use
>> case for that is to allow time intervals (eg for opening hours) to span
>> midnight, which are used in some countries, such as:
>>
>> Wednesday 18:00 – 25:00
>>
>>
>> However, it could also allow for the use of a term "midnight" for 24:00,
>> where that is the most natural expression.
>>
>> Mark
>>
>> Mark
>>
>> On Fri, Jan 22, 2016 at 7:40 AM, Martin J. Dürst <duerst at it.aoyama.ac.jp>
>> wrote:
>>
>>> In my opinion, "could be confusing" is a gross understatement :-(. I
>>> just recently wanted to submit some abstracts to a conference where I spent
>>> about 10 minutes to figure out which end of a day the actual deadline was.
>>>
>>> While there may be conventions for such things in some communities, and
>>> CLDR has an ambition to follow them, it's highly confusing in the world
>>> wide context of the web. The less such things are made defaults, and the
>>> more exact terms are used (e.g. "midnight at the start of the day" or some
>>> such), the better.
>>>
>>> Regards,    Martin.
>>>
>>> On 2016/01/22 10:30, kz wrote:
>>>
>>>> Dear CLDR users,
>>>>
>>>> I'm currently trying to implement in ICU the pattern characters b and B
>>>> for
>>>> datetime formatting, which involves the use of the word "midnight". See
>>>> http://www.unicode.org/reports/tr35/tr35-dates.html#Day_Period_Rules .
>>>>
>>>> Currently, according to the CLDR spec, the word "midnight" refers to
>>>> 0:00,
>>>> i.e. the beginning of the day. However, after a conversation with my
>>>> colleagues, we feel that it's more natural for "midnight" to mean, at
>>>> least
>>>> in English, the end of the day. For example, "Wednesday midnight" would
>>>> refer to midnight of Wednesday-Thursday, not the midnight of
>>>> Tuesday-Wednesday. This could cause confusion to users.
>>>>
>>>> In addition, other languages could have different problems with the use
>>>> of
>>>> "midnight". For example, Chinese has two different words for "midnight
>>>> (beginning of day)" (*lingchen*) and "midnight (end of day)" (*wuye*).
>>>>
>>>> As such, it'd probably be worth discussing to either (1) remove
>>>> "midnight"
>>>> as a time period, (2) use a different word for "midnight", or (3) modify
>>>> spec to have "midnight" refer to the end of the day.
>>>>
>>>> Any opinions?
>>>>
>>>>
>>>> Thanks
>>>> kz
>>>>
>>>>
>>>>
>>>> _______________________________________________
>>>> CLDR-Users mailing list
>>>> CLDR-Users at unicode.org
>>>> http://unicode.org/mailman/listinfo/cldr-users
>>>>
>>>> _______________________________________________
>>> CLDR-Users mailing list
>>> CLDR-Users at unicode.org
>>> http://unicode.org/mailman/listinfo/cldr-users
>>>
>>
>>
>> _______________________________________________
>> CLDR-Users mailing list
>> CLDR-Users at unicode.org
>> http://unicode.org/mailman/listinfo/cldr-users
>>
>>
>
> _______________________________________________
> CLDR-Users mailing list
> CLDR-Users at unicode.org
> http://unicode.org/mailman/listinfo/cldr-users
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://unicode.org/pipermail/cldr-users/attachments/20160122/79ad367b/attachment.html>