Thanks Philip Mountifield for the report and excellent analysis, and
Simon Charette for the review.

If settings.FIXTURE_DIRS contains duplicates or a default fixture
directory (app_name/fixtures), ImproperlyConfigured is raised.

Thanks to Berker Peksag and Tim Graham for review.

Refs #23663.

Previous versions of the tests were buggy, as initial_data.json
did exist and the test wasn't failing. It was finally failing on
Python 3.4.2.
Thanks Raphaël Hertzog for the report (and Debian bug #765117
contributors).

Thanks Rahul Priyadarshi.

Loading fixtures were failing since the refactoring in 244e2b71 for
inheritance setups where the chain contains abstract models and the
root ancestor contains a M2M relation.

Thanks Stanislas Guerra for the report.

Refs #20946.

`sort_dependencies` incorrectly interpreted 'complex' M2M relations
(with explicit through models) as dependencies for a model. This caused
circular complex M2M relations to be unserializable by dumpdata.

Thanks to aneil for the report and outofculture for initial tests.

Thanks tomwys for the suggestion.

Thanks Piotr Kasprzyk for help with the patch.

This was a regression in Django 1.6 that was only partially
restored in 839940f2.
Thanks Jonas Haag for the report.

Thanks Keryn Knight for the report.

Added ``--natural-foreign`` and ``--natural-primary`` options and
deprecated the ``--natural`` option to the ``dumpdata`` management
command.

Added ``use_natural_foreign_keys`` and ``use_natural_primary_keys``
arguments and deprecated the ``use_natural_keys`` argument to
``django.core.serializers.Serializer.serialize()``.

Thanks SmileyChris for the suggestion.

Tests by Tim Graham, report from jeroen.pulles@redslider.net.

Tests by Tim Graham, report from jeroen.pulles@redslider.net.

Should be unneeded with Python 2.7 and up.
Added some unicode_literals along the way.

It doesn't work as one might expect on a certain database backend where
autocommits_when_autocommit_is_off = True. That backend happens to be
popular for running tests.

This was a stealth option only used by the tests, and it isn't useful
any more since `atomic` provides nested transactions.

Fixed #20483.

Thanks Anssi for reporting this performance bottleneck.

If the fixture doesn't exist, loaddata will output a warning.

The fixture named "initial_data" is exceptional though; if it
doesn't exist, the warning is not emitted. This allows syncdb and
flush management commands to attempt to load it without causing
spurious warnings.

Thanks to Derega, ptone, dirigeant and d1ffuz0r for contributions
to the ticket.

The fixture named "initial_data" is exceptional though; if it
doesn't exist, the error is not raised. This allows syncdb and
flush management commands to attempt to load it without causing
an error if it doesn't exist.