Ross Barnowski

> Can I work on what you have suggested? Absolutely, please do!

This actually might be by design... I think the intended public interface for the VF2 algorithm is `nx.is_isomorphic`. If however there is an intention to document the `*Matcher` classes, it...

Thanks for taking a look @U-238. I think it'd be great if we could boil down these failures to the absolute minimum (i.e. what is the graph with the minimum...

I just stumbled across discussions in other projects on whether/how to adopt keyword-only arguments, e.g. scipy/scipy#14714. It appears scikit-learn has adopted this already and has a well-thought-out policy: https://scikit-learn-enhancement-proposals.readthedocs.io/en/latest/slep009/proposal.html FWIW...

Thinking about this a bit more, I think there are even more dimensions to consider (or: how to make the multiedge drawing at least as complicated as the single-edges!) For...

> This is just thoughts put down in email. Are there better ways to arrange this code? IMO yes, though the problem is not with the draw_multiedges function proposed here,...

> with relative import style users can simultaneously use several versions of the library by simply renaming it. like this True, but I don't think this would ever be recommended!...

IMO this is would be one of the most impactful features we could add to `nx_pylab`! I too like @ShirsenduP 's solution and it is likely to work well for...

Thanks for the report @harristeague , I can reproduce this as well.

This seems like it's related to #4895 . For instance, would we get the 10-15x speedup in `nx.selfloop_edges` mentioned in #4895 if `adj` were an attribute instead of a property?...