When converting to NetworkX, isolated vertices are dropped. Example:

M=gb.Matrix.from_edgelist([(0,1),(3,0)],nrows=5,ncols=5)
G=gb.io.to_networkx(M)

G.nodes gives NodeView((0, 1, 3)). Nodes 2 and 4 are both missing. This issue also affects viz.draw(). I do understand that it is one way to handle non-square matrices ...

Is this intentional or a bug? It is at least very surprising, and if intentional, it should be clearly documented.

Part of: pyOpenSci/software-submission#81