Ross Barnowski

> The combination of both shortest paths in final steiner graph would produce a cycle [t1, n1, n3, n2, t1]. That's an interesting example - do you happen to have...

Any thoughts on the latest comment @cw299 . Does fuzzing/contracting 0-weight edges fix the problem for your use-case? If this problem is really isolated only to graphs that have edges...

Agreed - this is a good idea. Another benefit of moving this is that the ThinGraph would actually be executable, which would serve as additional, practical testing of the factory...

> if the function doesn't call on core NetworkX functions or objects, then it probably shouldn't be included in the convert_matrix.py library of functions, is that right? No, that's not...

> The API surface area provides functions that return specialized data structures already. This is a good point. Looking at the other functions that are already in `convert_matrix`, there are...

I think the best way forward here is to generate a listing of the places where `is True` is used in the code, then evaluating them manually. FWIW we have...

The problem is that your nodes (which appear to be 2-tuple coordinates) contain whitespaces, which [is treated as the delimiter character by default for the `read_` and `write_` adjlist functions](https://networkx.org/documentation/latest/reference/readwrite/generated/networkx.readwrite.adjlist.read_adjlist.html)....

AFAICT this behavior isn't mentioned in the documentation anywhere, so the documentation seems like it could be improved here!

The read/write adjlist functions are very simple and don't really have a mechanism for generic node serialization (i.e. converting complicated node types to/from string formats). You might want to take...

> How about adding this text to the notes for write_adjlist? Thanks @brocla , your suggestion would do the trick IMO. Another option would be to add an example of...