Select and relocate work inconsistenly on an empty dataframe

[jttoivon]

I expect that trying to relocate a non-existing column with any_of would return the original dataframe. Even if the dataframe is empty. The function select seems to work more consistently with any_of.

My dplyr version is 1.0.7.

library(tidyverse)
tibble(a=1:2) %>% relocate(any_of("b"))
#> # A tibble: 2 × 1
#>       a
#>   <int>
#> 1     1
#> 2     2
tibble() %>% relocate(any_of("b"))
#> Error: Can't subset columns that don't exist.
#> x Location 1 doesn't exist.
#> ℹ There are only 0 columns.
tibble() %>% select(any_of("b"))
#> # A tibble: 0 × 0

^{Created on 2022-01-28 by the reprex package (v2.0.1)}

[DavisVaughan]

This does seem like a bug. Some related issues with .before and .after:

library(tidyverse)

tibble(x = 1) %>% relocate(x, .before = any_of("a"))
#> Warning in min(unname(tidyselect::eval_select(.before, .data))): no non-missing
#> arguments to min; returning Inf
#> Error in seq.int(from, to): 'to' must be a finite number
tibble(x = 1) %>% relocate(x, .after = any_of("a"))
#> Warning in max(unname(tidyselect::eval_select(.after, .data))): no non-missing
#> arguments to max; returning -Inf
#> Error in seq.int(from, to): 'from' must be a finite number

^{Created on 2022-01-28 by the reprex package (v2.0.1)}

I think we need to slightly tweak the logic of relocate() to better handle the case of empty selections. .before = <empty-selection> should probably be the same as .before = NULL, same with .after. This means if want to relocate columns before/after "nothing" then they end up at the front of the data frame, which is consistent with what relocate(data, cols) does.