Upcoming change to igraph's girth() function affects this package

[szhorvat]

Currently this packages checks if an igraph graph is a forest by computing it girth and comparing with 0:

https://github.com/mikldk/malan/blob/master/R/igraph.R#L26

 if (igraph::girth(x, circle = FALSE)$girth != 0L) stop("x must be a tree (or a forest)")

An upcoming version of igraph will change girth() to return Inf for graphs with no cycles (instead of the current 0). To keep compatibility with both the current and future versions, can you please compare both with Inf and 0?

This new version of igraph will also provide an is_forest() function, which will be slightly different than checking girth(): it will return FALSE for graphs with self-loops or multi-edges (which are ignores by girth()).

[krlmlr]

This package fails its checks for the release candidate for igraph 1.6.0. See https://github.com/igraph/rigraph/blob/main/revdep/problems.md for details.

Is this a problem in your package? Should we be doing things differently?

We're planning to release igraph 1.6.0 early December. This should give us at least until late mid-December, perhaps early January, to resolve this and keep CRAN happy at the same time. Thank you for your help!

Tracker: igraph/rigraph#968.

[mikldk]

@szhorvat @krlmlr Sorry for not attending to this earlier. It has now been addressed in 1829a9f and eca532b. My package is now compatible with both girth() returning 0 and Inf.

I will submit my update to CRAN on Thursday Nov 23.

[krlmlr]

Thanks. Friendly reminder: we're releasing igraph 1.6.0 in early December.

[mikldk]

Version 1.0.3 uploaded to CRAN: https://cran.r-project.org/package=malan. This should fix the issue so I close the issue. Thanks for good, downstream information.