- Remove `Page.get_expanded()`
- Postpone removal of `Page.edit()` since it's a more popular method.

per #130

Add OAuth and update docs

- Removing the need to use a double for loop to consume the the result
  by returning a list of revisions rather than a list of pages.
- Parsing the timestamps using the standard parse_timestamp method
- Adding tests

(#84)

Store the results of page.text() operations in a simple cache
dict. This avoids unnecessary remote roundtrips. Cache is
cleared on each successful page.save() operation. cache
argument can be set to 'False' to disable use of the cache.

- Moving code related to waiting/sleeping/retrying into a new class for a more object oriented approach.
- Removing any reference to wait "tokens" to avoid confusion with edit tokens.
- Note: `max_retries` and `retry_timeout` are no longer available on `Site`, but can still be passed into the constructor as before.

- Factor out parse_timestamp
- Use `iterkeys`, `itervalues`, `iteritems` and `next` from six
- Use __next__ for Python 3.x and next for Python2.x

Prepare for Python 3.x support
- Use print function
- Use compability methods from six

Back-ported to Python 2.6 and 2.7

- If database lag exceeds max lag, the API will still send 200, not 503.
  This was only added to documentation after bugzilla:31156.
  Ref. <https://www.mediawiki.org/wiki/Manual:Maxlag_parameter>
- In addition, Requests doesn't raise an exception when status code != 200
  unless raise_for_status() is called, so we should call that to catch
  any other 5xx errors such as Squid errors.
- Replace print statements with logging

Note that responses>=0.3.0 is needed to run the test suite now

As suggested in #52
Original auth mechanism introduced in #27
This change should retain backwards compability

Use new token handling system described on
https://www.mediawiki.org/wiki/API:Meta#tokens for MediaWiki >= 1.24

Work in progress, looks a bit dirty right now

- This is more flexible than the `Page.get_expanded()` method,
  since it can be combined with the `section` argument as well,
  to get the expanded wikitext of a section.

- Deprecating `Page.get_expanded()`.

- Also fixed a very minor bug where `Page.last_rev_time` was
  not reset when editing new pages after having edited non-new pages.

- Added more Page tests

since the method does not edit the page, but returns the text of the
page. The old method name still works, but issues a DeprecationWarning.
Note that DeprecationWarnings are silent by default, so we might
consider to elevate it in some future version.

Since we now require MediaWiki 1.16, these are no longer needed. It
might be necessary to add new compability methods in the future, but the
API is generally stable now.

- Require 'ordereddict' only on Python < 2.7
- Don't require 'simplejson'

- For some reason, 'pytest-cache' isn't found by setup.py if not
  specified explicitly in 'tests_require'
- Add tox.ini so tests can be run using tox if desired
- Update test to not be dependent on serialization

- A single command `python setup.py test` will now install test
  dependencies, make an in-place build and run the tests.
- The setup script now requires Setuptools. A message will be printed
  if it is not found.

- Note the following exception replacements:
  mwclient.errors.HTTPError -> requests.exceptions.ConnectionError
  mwclient.errors.HTTPStatusError -> requests.exceptions.HTTPError
  mwclient.errors.HTTPRedirectError -> requests.exceptions.TooManyRedirects

- Removed the `compability.old_upload` method, which was used for
  MediaWiki < 1.16.

- Requests automatically decompresses gzip-encoded responses, and
  does its best to decode response content to unicode when possible.

- [#53] Using the Responses library to mock http requests in tests

- Keep automated tests to be run with py.test in './tests/'
  and examples in './examples/'
- Rename test to adhere to 'Conventions for Python test discovery'
  (http://pytest.org/latest/goodpractises.html#test-discovery)