Hmm, interesting... now we need to figure out why these tests are failing.

It seems like regarding the Bron Kerbosch Max Cliques algorithm, the authors were of the impression that the algorithm works (in a consistent way) on both directed and undirected graphs. This is however wrong. Thus I adjusted the tests and docs accordingly.

Now the Docstring of Bron Kerbosch states that it can be called on directed graphs, but only if the directed graph is symmetric.

I also cleaned up the tests.

(The Steiner Tree Quickcheck(s) still fail and I'll take a look at that next)

Okay, similar to what was going on with Max Cliques, the Steiner Tree authors assumed that the algorithm would work on all undirected graphs (or at least did not specify that it does not). Instead, the algorithm (or rather the current implementation) only works on connected graphs. I adapted the quickcheck accordingly to skip graphs which are not connected.

I am not sure whether I like this solution, since a lot of arbitrary graphs should not be connected. Instead we could also call the steiner tree function on the biggest connected component. I would like some input on that :)

Also I cleaned up the Docstring again 🌞

I would like some input on that

It's not worth changing the implementation of steiner_tree right now, but in the next major release, we'll replace panicked operations with checked methods and return Result. Well, or we can refine the algorithm to steiner_forest, this is a debatable question - I'll open the issue a little later.

For now, lets limit to your documentation and quickcheck corrections. And please move all the changes regarding steiner_tree in a separate PR to make the change history more atomic.

Okay, I think I was not specific enough. I was asking for some input regarding the test setup we want to chose. Since currently we simply disregard all non connected graphs in the quickcheck, I was wondering how we could change that. I would edit the quickcheck to execute on all connected components (of size greater than 1) of the arbitrary graph and perform the checks there.

Regarding the separate PR, what do you mean by that? I only have Documentation changes in both the maximal_cliques and steiner_tree modules

I would edit the quickcheck to execute on all connected components (of size greater than 1) of the arbitrary graph and perform the checks there.

Oh, I guess I didn't read the question carefully. Yes, you can do testing for each connected component separately, if you wish. But even if disconnected graphs are simply discarded, the test will not lose much reliability, since the relative number of such input graphs is small. But in the future, this behavior will most likely be changed, and we will start checking that the algorithm returns something like Err(SteinerError::GraphDisconnected) on disconnected graphs.

Regarding the separate PR, what do you mean by that? I only have Documentation changes in both the maximal_cliques and steiner_tree modules

I just suggest moving the steiner_tree's docs and steiner_tree quickcheck's changes to new PR.

Regarding the separate PR, what do you mean by that? I only have Documentation changes in both the maximal_cliques and steiner_tree modules

I just suggest moving the steiner_tree's docs and steiner_tree quickcheck's changes to new PR.

Ah, I see 👍 And what about Maximal Cliques? Because I thought I'd do it all in one because otherwise the CI Will break if we update the quickcheck but don't update the quickchecks for Steiner tree and Maximal cliques

Ah, I see 👍 And what about Maximal Cliques? Because I thought I'd do it all in one because otherwise the CI Will break if we update the quickcheck but don't update the quickchecks for Steiner tree and Maximal cliques

I'll merge PR with steiner_tree fixes before that, so the CI won't break.
Regarding Maximal Cliques, I'll look at them today or tomorrow, I want to think over one of my ideas...

I'll merge PR with steiner_tree fixes before that, so the CI won't break.
Regarding Maximal Cliques, I'll look at them today or tomorrow, I want to think over one of my ideas...

Ahh, sry🤦‍♂️😅 I'll create a new PR then and also a new one for the maximal cliques changes 😁

Okay, I created two other PRs for Maximal Cliques and Steiner Trees respectively since I thought they'd both be split off anyways. See #800 and #801 . Also I reset this PR to only include the Quickcheck randomness change