TODO

For next version (0.8.5)
========================

* Add __init__

* Users should be able to add total and subtotals

* Missing lines to test with different versions of python in order to attain 100%
  code coverage:
  + Python 2.7: 11-16, 35-37, 65, 68, 73, 76, 199-202
  + Python 2.6: 11-16, 35-37, 65, 68, 73, 76, 199-202
  + Python 2.5: 11-16, 35-37, 65, 68, 73, 76, 199-202
  + Python 2.4: 11-16, 35-37, 65, 68, 73, 76, 199-202


For next version (0.9)
======================

* Apply some memoization algorithm so that results are not calculated each time the 
  attr is accesed. Since rows (the attr that brings the data to pivot) is a list and 
  both rows and result are attributes is pretty hard to apply some standard caché 
  algorithm: I need to find something that works with mutable objects, marking them 
  as dirty as soon some extra information is added to 'rows'. This gets even harder 
  since I want to support python>=2.4: as far as I can tell the most reasonable path 
  is:
  + for python>=2.6 use a mutable sequence type to perform some templating so that I 
    can detect when rows are dirty + find some way to hash my current rows to find 
    out if the cange is meaningful (monkeypatching the submitted object if it does 
    not define a __hash__, maybe?)
  + for python<2.6, just ignore the optimization part and let it work as it does now 
    or subclassing UserList and then follow the rest of the path suggested up supra.

* Add the possibility to have more than one to-be-pivotted column criteria (e.g. in 
  metrics, the result and the target for each month)

* Nothing else that I can think right now