Then there is the way standards such as ISO-8601 handle the problem... they ignore it, running the current calendar back, as if there was no calendar correction, into dim history. Such are called "proleptic" calendar, which I view as an artifical calendar not good for any dates, but an agreed upon way to handle the problem. People receiving such dates convert them to the local dates, as appropriate. I don't really LIKE this solution, however getting everyone to agree on what calendars changed when is very difficult. Especially since countries, such as the United States, changed on different dates depending on what country controlled the land at any particular moment. A real can of worms. --------------------------------------------------------------------- ghealton@exit109.com http://www.exit109.com/~ghealton/ ---------------------------------------------------------------------- Computers are like air conditioners: they don't work well when Windows are left open On Tue, 29 Jul 2003, Andrew Brown wrote:

...

...

it struck me that the tz data base would not be an unreasonable place to store the "local" date at which the gregorian reform took place

I privately started a little list to do that, but gave it up for lack of time. I found it hard to get authoritative data. It is hard to get reliable information about transitions that occurred that long ago.

i suppose something that was somewhat accurate (ie, culled from a few sources) as a starting point, and then was subject to readjustment (as necessary) wouldn't do? doesn't all the tz data work like that? :)

...

...

italy adopted the reform (1582/10/05),

That's true for most of Italy, but Toke Norby writes that Florence and Pisa didn't switch until 1749-12-21 Julian (1750-01-01 Gregorian).

...

...

or japan (1872/12/20).

You're giving a Julian date, but didn't Japan use a lunar calendar previously? And I've seen different sources about when the actual change occurred, ranging from 1873-01-01 Gregorian (your date) through 1919-01-01. Possibly they switched for some purposes at first but not others, which would mean there's no single date of switch.

actually, i was giving the date of the first skipped day on the julian calendar (so 1873/01/01 would be the first day actually occurring on the gregorian calendar) . since japan is "oriental", i can imagine them being subject to the footnote in calendrical calculations that notes that they don't use the gregorian calendar, but a reformed julian calendar. they're approximately the same...until 2800.

...

...

(modulo the various cantons in switzerland and the parts of germany that were protestant in the 16th century).

It's worse than that. In some of those places the date depended on who you were talking to. Even in the same town, part of the population would use Julian and the other part Gregorian.

yes, it might be difficult to decide on the granularity to use.

...

Also, once you go back that far, you should also address the issue of when the year started. For example, much of Europe started the calendar year on March 25 from the 12th through the 16th centuries. And some of Europe started the calendar year on Christmas Day.

yes, well, i was going to sort of ignore that, too. fwiw, i was also briefly beholden of the notion that the timezone database might hold "epoch" information, meaning that "second zero corresponds to calendar date yyyy/mm/dd" so that implementing a "hebrew" timezone (with the corresponding epoch) would be possible. then i read up on the structure of the hebrew calendar and decided to leave that to people who "need to know". if i "need to know", i'm sure i could find today's hebrew date much more easily than i could code it in an evening.

...

Anyway, if I haven't discouraged you from doing the job right (:-) here are some references that you may find of interest:

<http://www.norbyhus.dk/calendar.html> <http://www.tondering.dk/claus/calendar.html> Two good sources for general calendrical information.

got that from van vlijmen, but thanks.

...

<http://www.genfair.com/dates.htm> Explains calendrical issues for western Europe particularly well. Oriented towards genealogists.

i think i've found that one before. note, of course, that this is but one page, and the information is entirely duplicated in other "sources".

...

<http://www.calendarists.com> By far the highest-quality source of information for computerized calendars. However, they don't worry about transition dates much.

ah. more good work from the calendrical calculations people. very nice.

as a footnote, the shift from julian to gregorian calendaring (or something along those lines) is, imho, rather germaine to the tz data, since, for example:

% env TZ=Europe/Moscow date -r -1645700000 +"%B %e, %Y" November 7, 1917

as a sample date, is "wrong", because the localtime() routine ignores the gregorian reform. :)

-- |-----< "CODE WARRIOR" >-----| codewarrior@daemon.org * "ah! i see you have the internet twofsonet@graffiti.com (Andrew Brown) that goes *ping*!" werdna@squooshy.com * "information is power -- share the wealth."