Le 18/04/2012 22:14, Tom Lane a écrit :
> "Kevin Grittner" <Kevin.Grittner@wicourts.gov> writes:
>> I guess the point is that for hundreds of years, the same day could
>> have a different date depending which country's calendar you were
>> looking at. I'm not entirely clear why there's a problem if you
>> pick the Gregorian calendar and apply it retroactively.
> Which is, in fact, exactly what our code does. I think that bit in the
> docs is trying to explain why we do that rather than try to get the
> code to reflect what people really used back then.
OK, thank you both, that is clear for me know.
But maybe this sentence, without further explanations, is just a bit
confusing for a simple user, who will ask himself what to do with this
remark. If I understand it well, it has no consequences at all, except
that dates from 4713 BC are correctly handled by PostgreSQL.
Or maybe this sentence lacks only some linking words to explain more
clearly that the point is that PostgreSQL chose the Julian calendar
instead of other date conventions, because the other date conventions
are inconsistent before 1900.
Sorry, I am not fluent enougn in english to propose a patch.
>
> A possibly comparable point is that for timezone info we use the Olsen
> database (tzdata), which *does* make an effort to reflect historical
> realities. In consequence, at least once every several months we
> get somebody complaining about what a strange GMT offset he's seeing
> for timestamps before 1900 or so. If there's anyone out there who
> actually likes that behavior, we've not heard about it. (Not that
> I am going to try to get Olsen et al to change their policy.)
>
> regards, tom lane
>
Maybe this could be worth a small explanation somewhere on this page? (I
mean, explaining that before 1900, date and time can be inconsistent,
and lead to GMT offset that are not round).
