[osis-core] div type="date"

[Chris Little]

On Fri, 30 May 2003, Troy A. Griffitts wrote:

> That raises the question of whether or not we want date to allow a 
> 'general' value.  It's true we do have general, or 'container' values, 
> such as: 2003, 2003.01, etc.  These do not specify an exact time, but 
> rather a time narrowed down as far as is specified.  Does 'morning' mean 
> 'anytime between 06 and 12', or does it mean the precise time when the 
> sun came up on that day?
> 
> I still say use 06 and let the client software decide: if (06 and we're 
> talking about a 1st century hebrew conext) then display 'sunrise'.
>
> Other things come to mind like:
> 
> <div type="date" value="09.00">about the 3rd hour of the day</div>

These all identify too many significant digits.  01.01 should identify any 
hour/minute/second within January 1st.  01.01.T06 only matches identifies 
minutes/seconds within the 6th hour of Jan 1, not the span 
01.01.T00-01.01.T11, which is what 01.01.AM captures.

You could enumerate "01.01.T00 01.01.T01 01.01.T02 01.01.T03 01.01.T04
01.01.T05 01.01.T06 01.01.T07 01.01.T08 01.01.T09 01.01.T10 01.01.T11" as
the osisID, and that would at least constitute a correct encoding of the 
date for January 1st AM.  It's kind of long though.  But if we want to 
obey our own rules for spans and encode the date correctly, that's the way 
to do it for a date entry that matches part of a day.  (It would still 
inadequately describe sunrise/sunset though because that varies by time of 
year & latitude.  So if "moment of sunrise" needs to be encoded, it will 
have to be specified as something general like "sunrise".)
 
> Why are we using div?  And not <date>?

Completely different issue.

<date> marks occurrences in the text that record dates, like <date>January 
1st at 6 AM</date>.  What we're talking about here is <div>s of a calendar 
(daily devotional, lectionary, timeline, etc.) and what their osisID 
format ought to be.

--Chris