If we want to create what was provided prior to #7410 we could add another example like:

DG = nx.DiGraph(nx.dfs_edges(G, source=0))

But I think we decided that this approach did not produce uniformly random directed trees on pretty much any set of directed trees one might consider.

So, I put that example here in the PR instead of in the doc_string. A dedicated user who wants to reproduce the previous behavior badly enough can use that one-liner. But we're not going to advertise it for general (or new) users.

My only suggestion is maybe we should use G.edges inside nx.DiGraph.

Thanks @rossbar