Regression in dumping subelement in 0.13.3

[schlamar]

I'm not 100% sure if this is intentional or a regression, but there is different behavior in dumping subelements from 0.13.2 to 0.13.3. If this is not intentional behavior it is probably a regression from #419.

Test, passes on 0.13.2:

def test_tomlkit():
    a = tomlkit.parse('[a.b]\nk="v"\n[a.c]\nk="v2"')["a"]
    assert a == {"b": {"k": "v"}, "c": {"k": "v2"}}
    assert tomlkit.dumps(a) == '[b]\nk="v"\n\n[c]\nk="v2"'

Error Output on 0.13.3:

________________________________ test_tomlkit _________________________________

    def test_tomlkit():
        a = tomlkit.parse('[a.b]\nk="v"\n[a.c]\nk="v2"')["a"]
        assert a == {"b": {"k": "v"}, "c": {"k": "v2"}}
>       assert tomlkit.dumps(a) == '[b]\nk="v"\n[c]\nk="v2"'
E       assert '[a.b]\nk="v"\n[a.c]\nk="v2"' == '[b]\nk="v"\n[c]\nk="v2"'
E         
E         - [b]
E         + [a.b]
E         ?  ++
E           k="v"
E         - [c]
E         + [a.c]
E         ?  ++
E           k="v2"

[frostming]

Can you try to send a patch, I am not likely to work on this for now.

[schlamar]

Not really, sorry. Workaround would be tomlkit.parse(...).unwrap()