Possible to export `group_drops()`?

[earowang]

Possible to export group_drops() to achieve consistent behaviours with the dependent packages? Probably a better name instead of group_drops()?

[romainfrancois]

To add some context, group_drops() currently looks like this:

> dplyr:::group_drops
function(x) {
  !is_grouped_df(x) || is.null(attr(x, "groups")) || !identical(attr(group_data(x), ".drop"), FALSE)
}

and is used for e.g. the default value for .drop in group_by(). It probably should be a generic. I'm just not sure about the name at the moment.

[earowang]

It's probably too late to change the .drop argument to .preserve as in filter(), both !.drop and .preserve describing the same behaviour?

group_drops() is (and will be) used in the context of group_by() and filter()/slice()? maybe a more general name other than group_*?

[romainfrancois]

They don't exactly do the same thing :

• group_by(.drop=) is only about factors.
• filter(.preserve=) is about empty groups after the filtering, that can happen when grouping variables are non factors too.

[lock]

This old issue has been automatically locked. If you believe you have found a related problem, please file a new issue (with reprex) and link to this issue. https://reprex.tidyverse.org/