To-dos
======

Basic infrastructure
--------------------

   - Refactor SQL notebooks into structure and driver
   - Any obvious additional experiment combinators?
   - Convert JSON notebooks to use JSON-LD
   

Cluster functions
-----------------

   - Check handling of errors: are we masking too manye exceptions?
   

Testing
-------

   - Add test of notebook persistence, saving and coming back later


Documentation
-------------

   - UML class diagram
   