Datetime and Timestamp

[vng510]

Can you elaborate when a date will be stored as a python datetime object and when a date will be stored as a pandas Timestamp?
The reason why I ask is that it would be great to store everything in a datetime object so we don't have to worry about the Timestamp max limit of 2262-04-11. I see there is an option date_as_pandas_datetime but it says that if it's a true date, then it would be pandas datetime64 so does that mean that it will fall back to python datetime if needed?

[ofajardo]

By default a date is forced to be a python datetime.date object unless you set date_as_pandas_datetime in which case it is a pandas datetime type, pandas chooses it automatically, and keeps changing, so I am not sure what it is these days.

A datetime will be parsed to a pandas type, whatever pandas chooses it to be.

I did not understood what you are trying to do, but if you have dates or datetimes beyond the ranges of the default objects, you can set the option disable_datetime_conversion, then you get numbers that you xan convert manually to the object type of your preference.

A pandas

[ofajardo]

Also, I think you are mixing up datetime.date and datetime.datetime? A date may be stored as datetime.date , which max is the year 9999. So by default you are covered. A date will never be stored as a datetime.datetime. A date is defined as such in the source file sas, spss or stata.

A datetime is different, defined again at source and will be a pandas datetime/timestamp/whatever it wants it to be. Pandas is smart and you cannot force it to be a datetime.datetime (or it used to be like that at least)