"Mailing List" <li...@fastmail.net> writes: > Was including a input check on a function argument which is expecting a > datetime.date. When running unittest no exception was raised when a > datetime.datetime instance was used as argument. Some playing with the > console lead to this: > >>>> import datetime > >>>> dt1 = datetime.datetime(2010, 10, 2) >>>> type(dt1) > <type 'datetime.datetime'> >>>> isinstance(dt1, datetime.datetime) > True >>>> isinstance(dt1, datetime.date) > True > >>>> dt2 = datetime.date(2010, 10, 2) >>>> type(dt2) > <type 'datetime.date'> >>>> isinstance(dt2, datetime.datetime) > False >>>> isinstance(dt2, datetime.date) > True > > My issue (or misunderstanding) is in the first part, while dt1 is a > datetime.date object (confirmed by type), the isinstance(dt1, > datetime.datetime) returns True. Is this correct? If so, is it possible > to verify whether an object is indeed a datetime.date and not a > datetime.datetime object?
I think you got it wrong here: dt1 is a *datetime* object, and also a date-object. I guess that's because it conforms to the same protocol - it has year, month and day and so forth. > > For background information; reason my choice was to use the > datetime.date object is that there should not be ay time information in > the object. Of course this can easily be stripped off in the function, > but thought the use of a datetime.date would be a nice shortcut... Obviously not :) You can of course test not on inheritance (which isinstance does), but actual type equality: >>> type(dt1) == date False Diez -- http://mail.python.org/mailman/listinfo/python-list
