Adjust "null" to expected behavior

[kalzoo]

Add "not" to ensure that when null is True (making None or '' distinct from False) that the default placeholder (--) is rendered

[jieter]

@kalzoo your change make 4 tests fail. Can you have a look at that?

[kalzoo]

@jieter, all four appear to assert the behavior that I believe is unexpected. I think it's just a semantic thing, but I would expect:

null=False  # to mean that null is *not* a distinct value
null=True  # to mean that null is a distinct value from true and false, 
                  # consistent with how the Django ORM uses it.

However, the opposite expectation is reflected in your tests, for example (tests/columns/test_booleancolumn.py:44):

    def test_treat_none_different_from_false(self):
        class Table(tables.Table):
            col = tables.BooleanColumn(null=False, default='---')

        table = Table([{'col': None}])
        assert table.rows[0].get_cell('col') == '---'

In my mind, "none different than false" should mean null=True rather than False.

Edit re: ORM

[jieter]

@kalzoo thanks for fixing the tests, however, this is a breaking change and this PR provides no deprecation path.

Since this is just an issue of expectation, can we maybe also adjust the documentation (docstring) to make sure future expectations are correctly created?

[kalzoo]

True. Seemed low-impact since it won't cause exceptions, only a minor difference in output.

As a new user to django, this change is more intuitive to me, especially since it would make null usage consistent with django ORM/models. But that may not be the case for longer-time users, so it's just up to your discretion.

In any case, the library's great to have, thanks for your work in maintaining it.

[jieter]

Thanks for your thoughts, I think I'll take this into django-tables2==2.0, refs #526