Please refer to igraph instead of python-igraph

[szhorvat]

Please refer to igraph instead of python-igraph in the following location:

https://github.com/netZoo/netZooPy/blob/master/requirements.txt#L6

and remove this entry, since igraph is already included:

https://github.com/netZoo/netZooPy/blob/master/setup.py#L16

Please see igraph/python-igraph#699 for an explanation.

---

Additionally, the igraph-related troubleshooting section has been obsolete for a while and can be removed. Binary wheels are made available for all common platforms.

https://github.com/netZoo/netZooPy/blob/master/docs/install/index.md?plain=1#L36

---

Finally, I would strongly recommend not constraining the igraph version to before 0.10, as you are missing out on many bugfixes. I see that this was done to ensure that some functions keep returning the same cluster assignment. I am not sure which functions you are referring to, presumably some randomized ones. Keep in mind that generally there is no single correct cluster assignment, which is precisely why many of the algorithms are randomized. To get an accurate picture about clustering, you need to run the algorithm multiple times, and see if the result is reasonably stable. Picking just one result is not a scientifically solid decision ...

[violafanfani]

Thanks for your help.
This has been fixed with PR #316.

We updated the documentation and now avoid constraining to a specific igraph version.
In the future we will need to rethink our unit tests such that they are not so dependent on the package version.

[szhorvat]

Feel free to ask us for advice if there's a concern about the best way to test functionality that relies on stochastic igraph functions.