Date: Wed, 2 Mar 2011 10:11:47 +0100
Subject: Re: Requirements // Ordinal and Week dates
On Mon, 28 Feb 2011 11:40:50 -0500, Ray Denenberg, Library of Congress wrote
> Once again I ask for an explanation of the requirement.
> I have two areas of confusion.
> First, the recent examples include, for example, "3rd Monday in Feb. 2011".
> Is this a real example? The third Monday in February of 2011 is a real
Why not. That date is not Washington's Birthday. His birthday is generally
given asdate is 22 Feb. We can't even speak of President's Day as that is
not its name in many states. It is also a bank holliday in Canada and they
do not even have Presidents :-)
> and can be represented as "2011-02-21". I had thought that you wanted to
And if it was that bank holliday in one of 2009, 2010 or 2011?
And while we are at it why not add moon to the set analogous to week + day?
Throw in alongside "intervals" (or precisions) for Seasons the Equinoxes,
Maxima (Mid-summer Soltice) and Minima (Mid-winter Soltice) we can pick up
quite a few named dates without using their cultural names..
If we use Month names Sept (09 or some other code) and March (03 or ..)
instead of Autumn and Spring we can even sidetrack the Hemisphere problem.
> represent "third Monday in February", independent of year, in order to
> represent a pattern, as in this case a particular US holiday, i.e. George
> Washington's Birthday, falls each year on the third Monday in February.
> Second (as Bruce has already asked) assuming you want to represrent "third
> Monday in February" why can't it be represented as the human readable
> 'third Monday in February', or '3rd Mon. in Feb'; i.e. why does it need to
> be normalized for computer readability? Can you provide a use case where
> machine readability is necessary?
I would like to have these for search and retrieval application-- for
example within SRU/W. I do not see them as being used to mark a date in a
record but as values for date in a search...
Give me the news that occured on U.S. Nov election day in 1984?
This could be quite useful in canned queries, facets, etc...
20uu-11-D12 are the first Tuesdays of November in our "century".
A canned search facit for uuuu-11-D12 can be interesting.. just as for 19uu-
05-01 (May Days in the 20th century)
There are other interesting dates.. such as Easter.. Easter pogroms or
Some days are symbollic on fixd places in the calendar.. such as the Roman
Catholic Christmas, May Day, Guy Fawkes Day.. some even are named after
their date: 5 May, 4 July, 11 Sept etc.
Others are fixed by other rules such as the U.S. Labor Day (first Monday in
Sept to mark a contrast with 1 May which comemorates the Chicago Haymarket
Riot of 1886), Easter, Newroz, Festival of Eid-ul-Fitr (the end of the month
I do not think it would be wise to try to create a controlled vocabulary for
these dates. They are too loaded and we don't want to open this Pandora's
These constructs can be quite useful for searches.. A query to search a
databse for accident reports on the U.S. Labor weekends in the 1980s can be
easy to formulate with an appropriate means to express it or difficult
On Fri, 3 May 2013 13:33:28 -0400, Christopher Weeg wrote
> Hello, all --
> My name is Chris and I just discovered this Extended Date/Time Format
> group. I'm a recent library school graduate.
> I just read the draft specification, which covered many of the permutations
> that I have been thinking of over the last year or two. Very cool.
> One other thing that I've thought of and did not *notice* being covered in
> the draft spec:
> I wonder if there is a way to represent an algorithmically repeating date
> event? The first examples that come to mind are elections, like the US
> presidential election occurs once every four years (subsequent to some
> originating date, presumably). Considering the algorithm (first Tuesday in
> November?), days of the week would need to built into EDTF, I suppose.
> There's also the Olympics. Or how about Easter (in which case one would
> need to incorporate celestial phenomena, i.e. the vernal equinox and the
> occurrence of a full moon...?). You have to re-up your car insurance every
> 6 months. Etc., etc.
> Anyway, just a thought. I searched the listserv history for discussion of
> this topic and didn't find anything. Not sure whether this topic even fits
> within the scope of EDTF, but once you start getting into discussing
> intervals and multiple dates (5.3.4) it seems like maybe repeating/cyclic
> date events are not too far behind...
> Thanks for reading, and for any suggestions or comments you might add.
> -- Chris Weeg
Edward C. Zimmermann, NONMONOTONIC LAB/BSn