Dan Schult

It looks like this PR has a few extra files from isomorphism. These are probably additions you were making to update the Pr to match the main branch. I think...

If the layout functions create a dict/array of positions during the algorithm and then stores it on the graph it might be good to return the dict in addition to...

> I've left a warning block in the documentation about the default value of attribute changing in the future. It feels like more than just a change in the default...

One way to make the release cycles 3 or 4 instead of 2 would be to merge this PR with the deprecation warnings removed. Then 3.4 has both functions without...

> If we decide to wait a cycle or two before adding the deprecation warning, I strongly think that we should add warnings in 3.4 when the new draw API...

I haven't dived into this very far. But my initial reaction is that individual functions already cache graph properties and even data structures. We should do that rather than change...

The cached_property is already used in NetworkX for the e.g. `edges` property IIRC. So that might be a way to go. But I think we decided to only do that...

To see if this is worthwhile we could check the timing of running the entire test suite with this branch and the main branch. We could also look at the...

Dispatchable is actually called before not_implemented_for... The order is not obvious, but the "inner" decorator goes first IIUC. For example, ```python @not_implemented_for("directed") @nx._dispatchable def bridges(G, root=None): ``` This makes the...

I don't see changes on the scale of nanoseconds. Am I looking in the wrong place (the shortest path benchmarking PR #8059 ) It seems like some cases are ms...