Ideas ===== A collection of some random ideas & thoughts for future implementations * Easier access for the `through` table for :py:class:`django.db.models.ManyToManyFields` than requiring the user to manually access ``Model.column._through()`` * We could use SQLAlchemy to generate the SQL for :py:func:`django.db.connection.cursor`, allowing reuse of Django's current transaction. This should currently work, but we could add some sugar for making it easier & documenting it. It would mean that you no longer get a SQLAlchemy ResultProxy back.