* Performance
* Parallel df.evaluate [474](https://github.com/vaexio/vaex/pull/474)
* Avoid calling df.get_column_names (1000x for 1 billion rows per column use) [473](https://github.com/vaexio/vaex/pull/473)
* Slicing e.g df[1:-1] goes much faster for filtered dataframes [471](https://github.com/vaexio/vaex/pull/471)
* Dataframe copying and expression rewriting was slow [470](https://github.com/vaexio/vaex/pull/470)
* Double indices columns were not using index cache since empty dict is falsy [439](https://github.com/vaexio/vaex/pull/439)
* Features
* multi-key sorting of a DataFrame [463](https://github.com/vaexio/vaex/pull/463)
* vaex expression to pandas.Series support [456](https://github.com/vaexio/vaex/pull/456)
* Dask array support [449](https://github.com/vaexio/vaex/pull/449) [#476](https://github.com/vaexio/vaex/pull/476) [example](http://docs.vaex.io/en/latest/example_dask.html)
* isin() method for expressions [441](https://github.com/vaexio/vaex/pull/441) [docs](api.html#vaex.expression.Expression.isin)
* Existing expressions are rewritten, to make them behave like arrays [450](https://github.com/vaexio/vaex/pull/450)