Re: CLDR proposal: Holiday rules

From: Stephen Colebourne (scolebourne@btopenworld.com)
Date: Fri Oct 21 2005 - 07:45:46 CST

  • Next message: Peter Constable: "RE: Thai v. Lao (was: Lao Letter Fo Sung and Lao Letter Fo Tam)"

    I am at work at present, and don't have time to
    respond to all these good points. However....

    > 7) Since it's hard to find a design document or the
    > DTD of this subject on
    > the sourceforge.net site, I'll make a guess that the
    > "year" attribute
    > means that holiday happens once
    Yes

    >, while "fromYear" happens more than once.
    > Is there a "toYear" attribute?
    Yes

    I attached the DTD to the original mail, but I don't
    think it got through. See
    http://cvs.sourceforge.net/viewcvs.py/joda-time/JodaTimeContrib/holiday/src/java/org/joda/time/contrib/holiday/src/
    for the CVS DTD. Please bear in mind that this CVS is
    essentially a whiteboard area at this point, and there
    is no fomal design process.

    > 8) Instead of using things like "Wales" for an
    > identifier, have you
    > considered using ISO-3166-2 or even LOCODE from the
    UN?
    I didn't know there were region codes! Looks v useful
    for this task.

    The identifier is a key and difficult aspect. Time
    zones have settled on Continent/City, such as
    Europe/London, but holiday data seems trickier to
    classify than this...

    Stephen

    --- George Rhoten <grhoten@us.ibm.com> wrote:
    > I have a few comments on this information.
    >
    > 1) A lot of this data seems to be very Gregorian
    > Calendar oriented. What
    > about other calendars, like the Julian, Chinese,
    > Hebrew and Islamic
    > calendars? For example, Christmas isn't celebrated
    > around the world on
    > the same day. The Chinese New year is based on the
    > Chinese calendar.
    >
    > 2) How would Indian weekends be handled? The
    > weekend data is already in
    > CLDR, but CLDR doesn't quite handle Indian weekends
    > that are variable in
    > length depending on the time of the month.
    >
    > 3) I really recommend that you don't use the term
    > "country" in any of your
    > elements. You should consider using something like
    > "region" and
    > "subregion", or "territory" and "subterritory". Not
    > all ISO-3166 codes
    > are countries. Implying that all ISO-3166 codes are
    > countries tends to
    > annoy certain country governments controlling
    > certain territories. It
    > implies that some territories are self governing.
    >
    > 4) You may want to add information about the level
    > of vacation or
    > observance. Some holidays are observed, and others
    > are taken as
    > vacations. For example, banks and schools stay open
    > and celebrate on St
    > Patrick's day in the US, but almost everything
    > closes on Christmas day.
    > This is a fuzzy concept, and it may be hard to
    > quantify.
    >
    > 5) I don't quite get the concept of baseRegion. Is
    > this a way to do
    > inheritance?
    >
    > 6) How are multi-day holidays handled? For example,
    > there is the month of
    > Ramadan and the 40 days of Lent. These aren't
    > traditional holidays, but
    > these are important periods of time that are
    > observed.
    >
    > 7) Since it's hard to find a design document or the
    > DTD of this subject on
    > the sourceforge.net site, I'll make a guess that the
    > "year" attribute
    > means that holiday happens once, while "fromYear"
    > happens more than once.
    > Is there a "toYear" attribute? For example,
    > Thanksgiving day in the US
    > got moved several times
    >
    (http://www.loc.gov/wiseguide/nov02/thanks-when.html).
    > While Thanksgiving
    > day is uniform at the moment, other holidays may
    > change like this in the
    > future.
    >
    > 8) Instead of using things like "Wales" for an
    > identifier, have you
    > considered using ISO-3166-2 or even LOCODE from the
    > UN?
    >
    (http://www.iso.org/iso/en/prods-services/iso3166ma/04background-on-iso-3166/iso3166-2.html)
    >
    > While it doesn't cover all countries around the
    > world, it may be
    > something to think about.
    >
    > 9) The concept of state is a blurry one. Instead of
    > using type="State",
    > wouldn't type="government" be better?
    >
    > Overall, it's all an interesting and tough problem
    > to solve.
    >
    > George Rhoten
    > IBM Globalization Center of Competency/ICU San
    > José, CA, USA
    > http://www.icu-project.org/
    > http://icu.sourceforge.net/



    This archive was generated by hypermail 2.1.5 : Fri Oct 21 2005 - 09:33:47 CST