TODOs in `socnet.R`

[maelle]

There are a few TODOs in socnet.R but it made me wonder why igraph has this interactive interface. Should it still be supported?

[szhorvat]

IMO interactive plotting and graph editing for small graphs (tkplot()) is very useful. We invested in creating a similar but more polished and more usable interface in Mathematica, and many users appreciate it. Perhaps the R version just needs more advertisement, or some improvement. What's holding it back is no support for adding/removing vertices/edges, which would be the the most useful possible feature ...

As for the analysis UI (tkigraph()), I am much more skeptical about this. It might make more sense to allow other GUI tools to integrate well with igraph rather than maintain this ourselves.

Both of these are my personal opinion, and I expect others may disagree.

[szhorvat]

It'd be good to try to get an idea about practical usage before removing any of this functionality. Suppose someone uses tkigraph() for teaching, but we don't even know about it.

[krlmlr]

I don't see evidence of tkigraph() being used. Can we move this part to a separate package on GitHub?

https://github.com/search?q=tkigraph+language%3Ar+-path%3A**%2Fsocnet.R++-path%3A**%2Ftkigraph.Rd++-path%3A**%2Ftkigraph.html&type=code

[krlmlr]

tkplot() seems to be used.

https://github.com/search?q=%2Ftkplot%5B%28%5D%2F+language%3Ar+-path%3A**%2Ftkplot.R++-path%3A**%2Ftkigraph.Rd++-path%3A**%2Ftkigraph.html&type=code

[krlmlr]

Is there a more modern HTML- or JS-based alternative for tkplot() ?

[maelle]

If we remove tkigraph() in a release and someone complains, at least we'd know it's used? 😅 (I obviously don't want to willingly break anyone's teaching materials)

[szhorvat]

Factoring out tkigraph to a separate package (which might receive minimal maintenance for a little while) sounds like the best idea. It doesn't even need to go on CRAN. This way:

• We do hear complaints if someone uses it.
• We do have an immediate solution for people who complain.
• Support can be dropped completely if there are no complaints.

We did the same with the spectral coarse graining functions in the C core, https://github.com/igraph/igraph-scg Keeping them functional after igraph updates turned out to take near-zero work.

[github-actions]

This old thread has been automatically locked. If you think you have found something related to this, please open a new issue and link to this old issue if necessary.