--------------------------------
- Tab completion of all Flickr API functions in IPython.
- Fix for a bug where a crash occurred when XML containing
a `<name />` element was parsed with XMLNode.
- Deprecated a number of methods for old-style error handling, including the
`fail_on_error` constructor parameter. Just handle the `FlickrError` exceptions
instead of explicitly testing all method calls. The deprecated methods will be
completely removed in release 1.2.
- Implemented a response parser system, which still parses to XMLNode objects by
default. It now also includes parsing to the Python standard ElementTree, which will
soon replace XMLNode as the default response parser. See the
[documentation](http://flickrapi.sf.net/flickrapi.html) on how to use the new parsers.
- Added `format` constructor parameter to set the default response format for all
method calls. Overriding per call is still possible.
- Added `store_token` constructor parameter that's `True` by
default. Set to `False` to ensure the on-disk token cache isn't
used. [[ Feature request 1923728 ]](http://sourceforge.net/tracker/index.php?func=detail&aid=1923728&group_id=203043&atid=984009)
- Added caching framework. Pass `cache=True` to the constructor to
use it. [[ Feature request 1911450 ]](http://sourceforge.net/tracker/index.php?func=detail&aid=1911450&group_id=203043&atid=984009)