Loading Tasks ============= Graph generation involves creating tasks for each kind. :doc:`kind` are ordered to satisfy ``kind-dependencies``, and then the :term:`Loader` specified in ``kind.yml`` is used to obtain the initial set of tasks for that kind. In short, loaders are responsible for turning the ``kind.yml`` file into an initial set of tasks that can then be passed down into the :term:`transforms `. What is a Loader? ----------------- Loaders are Python functions with the following signature: .. code-block:: python def loader(cls, kind, path, config, parameters, loaded_tasks): pass ``kind`` is the name of the kind. ``path`` is the path to the configuration directory for the kind. This can be used to load extra data, templates, etc. ``config`` is a dict containing the contents of the ``kind.yml`` file. ``parameters`` give details on which to base the task generation. See :doc:`/reference/parameters` for details. ``loaded_tasks`` contains all tasks that have already been loaded (which is guaranteed to contain all kinds of tasks listed in the ``kind-dependencies`` key). The return value is a list of inputs to the transforms listed in the kind's ``transforms`` property. The specific format for the input depends on the first transform - whatever it expects. Available Loaders ----------------- Taskgraph provides two loaders, although in truth they are virtually identical: The :mod:`Transform Loader` is a barebones loader. The :mod:`Default Loader` does everything that the Transform Loader does, and additionally ensures that the :mod:`Run ` and :mod:`Task` transforms are used on every task it loads. Projects may optionally provide their own loaders that support different behavior.