Model classification
====================

Model classification works via the collections
feature. Each different classification scheme used
should have its own root node. Each subsequent node
in this classification must use a node type that
has the ``is_classification`` flag set to ``True``.

You can add any number of ``Attribute``\s to a node,
which will then be passed on to all linked model instances.

**Example:**

You have a model named *Product*, it has two attributes:

  - name: `str`
  - color: `str`

Now you can create any number of *Products* in your master data.

Create a new classification collection using any number of nested
nodes; usually only leaf nodes will have attributes.
Now if you add an attribute to a classification node:

  - price: `decimal`

And you link your created model instances of *Products* to this
classification node, every Product in this node will have the new
attribute `price` available.

In this case the API response for a single master data item
will include a list of linked classification names (using the root node)
and a list of attribute IDs as specified in the linked nodes::

    {
      "classifications": {
         "eCl@ss 9.1": [
            "57c03cb2a69a528ccc4862c4",
            "57c03cb2a69a528ccc4862db"
         ]
      },
      data: {
        // ... etc
      }
    }

The values for these additional attributes are all saved together
in the primary ``fields`` dictionary, as duplicate attribute keys
will simply share their value for convenience.