group_if, group_at, group_all

[krlmlr]

group_by(everything()) doesn't work, probably because group_by() has mutate() semantics.

group_by_all <- function(.data) group_by_(.data, .dots = colnames(.data))

This is different from rowwise() iff the data contains duplicate rows.

[hadley]

I wonder if we should just provide a variant with select() semantics. group_by_select(), group_by_where()?

[krlmlr]

set_group_vars() and add_group_vars()? Simply group()?

[hadley]

I think it needs to have a similar grammatical form to group_by()

[hadley]

It now seems obvious that this should be group_by_at(), group_by_if(), group_by_all(), and so I'll give it to @lionel- 🙂