The lax handling of nulls in the JSONRdfWriter and anzo.js reader (and presumably the writer and reader going the other direction as well) can make debugging difficult.
This was particularly true for a recent bnode issue where BlankNodes created with a different BlankNodeManager than used by the JSONRdfWriter were serialized to null because the JSONRdfWriter was unable to find a label for the BlankNodes since it's manager had not created them.
This is how the null blank node looks on the wire:
... subject":{"objectType":"bnode","value": null } ...
...
... object":{"objectType":"bnode","value": null } ...
On the anzo.js side, the null valued blank node was not rejected (no exception or error was thrown), but it was not correlated with the other uses of the same blank node within the graph either.
The result is that in the constructed anzo.js graph, a single blank node appearing in a number of statements was treated as numerous, unrelated, blank nodes.
