Leap years didn't exist prior to AD 45, and then leap centuries were added from 1600 onwards (and 11 days missed out in 1582 to compensate for the leap years that were added on centuries that shouldn't have had the extra day added).

If the astronomical calendar is the same as Gregorian all the way to year 1, then it must handle those rules the same as the Gregorian calendar, so there shouldn't be any leap years prior to AD 45 at all.

Although that said, the month lengths were different before AD45 and there was sometimes an additional month inserted to realign the months and seasons. So trying to apply Gregorian calendar rules prior to AD45 seems a bit of a fool's errand anyway.

If the astronomical calendar is used to make calculations easier (e.g. replacing 1BC with 0 and 2BC with -1), I'm not sure that they'd go to the effort of trying to insert additional months before AD45 just to line up with the Gregorian calendar (which it doesn't before AD1 anyway), so it'd make sense to simply use our current months and not apply any leap year logic prior to AD45 and just not worry about attempting to line up any dates prior to AD45 in the calendars at all.

And of course, any historical documents prior to the fifth century would have been using a different calendar system for the years anyway, and would have to have been converted to AD or BC, so there's even less reason to worry about trying to make the dates in their calendar match up with the same exact dates with the modern calendars.

https://www.postgresql.org/docs/9.0/functions-datetime.html#...