v0.1.0, 31/01/2014 -- Initial release
v0.2.0, 31/01/2014 -- search method now accepts empty params and takes default filters and attributes when available
v0.3.0, 05/02/2014 -- added tests and fixed some bugs
v0.4.0, 27/02/2014 -- using requests' stream = True on GET method when retrieving response
v0.4.1, 24/11/2014 -- fixed "KeyError: 'displayName' when calling BiomartDataset.search"
v0.5.0, 25/11/2014 -- design change: not using params dict to initialize classes, using **kwargs instead
v0.6.0, 04/06/2015 -- fixed issue with using the 'default' virtualSchemaName. Now fetching it from the registry. Using default attributes from the first visible attributes page to prevent a biomart exception from being raised (cannot use default attributes from multiple attribute pages).
v0.7.0, 09/09/2015 -- should be usable under a python3 env
v0.7.1, 09/09/2015 -- fixed relative imports for python3
v0.8.0, 01/10/2015 -- New design of classes, new way of fetching attributes and filters for datasets. Still need to figure out the actual need for virtualSchemaName, for now 'default' is used for all datasets.
v0.9.0, 07/04/2016 -- The module is now aware of AttributePages and will check whether selected attributes all belong to the same page. thanks to krassowski for his work.
v0.9.2, 22/01/2017 -- Now considering virtualSchemaName value when searching (was always using 'default')