TODO

Ongoing
=======

[ ] Keep unittests and nose coverage up to date

[ ] Keep documentation up to date

Alpha 
=====

[x] Implement basic Mongo stuff

[x] Implement Elastic stuff

[x] Implement RESTful views (for CRUD, "forms" and search)

[x] Setup a pcreate scaffold

[x] Flesh out support for inter-object references and be sure to link
    to referenced objects in "_links" section of object GET view
    (with a new namespaced "reference" rel)

[ ] Create a utility for keeping references on the other end of
    many-to-many relationships up to date when a context object is saved.

[ ] Implement HierarchicalCollection and FolderObject

[x] Add support for files and GridFS

Maybe?
======

[ ] Implement some sort of auto-naming collection and/or object subclasses.
    Such collections should use POST instead of PUT for new object creation
    (like the base Collection).

[ ] Rethink use of request throughout resource classes... can we ditch
    it?  grep request resources/*.py
    *** On the other hand... having access to the request would be handy 
    for things like logging the user ID for certain actions... hmm.
    [ ] Change root_factory and Root so that request.registry.settings is
        passed to Root intead of request (for access to Mongo, Elastic, etc).
    [ ] Change Object.get_class_schema() such that it doesn't take a request
        arg and instead uses a schema defined by a class attribute.
        Change Object.get_schema() such that request is required, but the
        default implementation just returns get_class_schema().  Subclasses
        that need to customize the schema based on request can do so in a deepcopy.
    [ ] change File such that it's passed the gridfs instead of request
    [ ] update tests
    [ ] update scaffold
    [ ] update docs!

[ ] Index files in Elastic using its support for attachments

[ ] In REST views, treat files as separate resources.
    Example: /people/sam-brauer/photo (GET, PUT, DELETE)
    If an object has a list of files, support POST too:
    Example: /galleries/robots/photos/

[ ] Take a closer look at pyelasticsearch as an alternative to pyes

[ ] consider a simpler htmlutil.html_to_text() that uses lxml like so:
    lxml.html.fromstring(s).text_content()
    (at the expense of adding dependency on lxml)

[ ] replace is_html attribute with a text_type attribute...
    with values 'plain' (default), 'html', 'markdown', etc