Svein Erling Tysvfr wrote:

> I know that one in theory does learn from experience, but storing time
> history tables for trains further back than the year 100 is
> ridiculous. Or is British Rail (or whatever they're called now)
> storing them because they want to say "Hey, our trains used to be 1800
> years delayed, now you only have to wait a few hours - so stop
> complaining!"

Given the performance it's not too unreasonable, but if you have been
following any other threads you will see I have been working on
genealogical data for the last couple of years.

> Nor does it sound sensible storing exact dates before year 1 (trying
> on Firebird 1.0, I found that minimum date was 0001-01-01, not
> 0100-01-01 like you wrote), we don't know what date Jesus was born
> (even though we celebrate it 25 December) and the celebration of his
> death/resurrection varies from year to year (we do know that it was
> two days between his death and resurrection). And I'd be greatly
> impressed if the Caine family or anyone else kept a family record more
> detailed than that. Hence, I think an integer field for year or
> possibly for month if you want to be very detailed would be more
> appropriate for dates this far back in time.

The QUOTED minimum is January 1, 100AD ( page 945 of the book ) which
would correspond to a correct/incorrect leap year in the Gregorian
calendar, which is one reason for asking the question.
*I* currently have dates back to 1000 AD, which are on the Julian
calendar, and can be converted to Gregorian day counts for comparison
purposes, with the period from 1500 to 1900 needing to converted
depending on the country and date, to provide a normalised day count.
While we do not probably need day accuracy in roman times, despite the
mess that Old Julius sorted out in 45BC there are documented dates back
into that period, and Astronomical calculations go back to a day 0 of
1 January 4713 BC ( Midday since astronomers work nights ;) )

Working from this basis, all of the other calendars used world wide can
be overlayed with matching days. If you were Jewish you would know that
the world was created in 3761 BC somewhat after day 0 so can still be
calculated, and Jewish records preceed the birth of Jesus.

*SO* Do I use a simple DATE or TIMESTAMP and adjust for variations, or
do I throw them in the bin and start again with a REAL date element as
an integer starting on day 0 :)

--
Lester Caine
-----------------------------
L.S.Caine Electronic Services