Still cannot mutate a grouping column if using across()

[DanChaltiel]

Hi,

Since #4709 and version 1.0.0, mutate() and summarise() allow modification of grouping variables, which is very useful.

However, while this works nice and clean using the standard syntax, I get an error when I try to use across().

Indeed, as said in #5963, ?across() tells us that:

Because across() is used within functions like summarise() and mutate(), you can't select or compute upon grouping variables.

This feels a bit inconsistent, and if it is still a real limitation today (I hope not), I guess at least that the justification in the doc is inadequate.

Here is a little reprex anyway (I use the dev version of dplyr) :

library(tidyverse)
mtcars %>%
    head() %>% 
    group_by(cyl) %>%
    mutate(cyl=sum(cyl))
#> # A tibble: 6 x 11
#> # Groups:   cyl [3]
#>     mpg   cyl  disp    hp  drat    wt  qsec    vs    am  gear  carb
#>   <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
#> 1  21      24   160   110  3.9   2.62  16.5     0     1     4     4
#> 2  21      24   160   110  3.9   2.88  17.0     0     1     4     4
#> 3  22.8     4   108    93  3.85  2.32  18.6     1     1     4     1
#> 4  21.4    24   258   110  3.08  3.22  19.4     1     0     3     1
#> 5  18.7     8   360   175  3.15  3.44  17.0     0     0     3     2
#> 6  18.1    24   225   105  2.76  3.46  20.2     1     0     3     1
mtcars %>%
    head() %>% 
    group_by(cyl) %>% #removing this line removes the error
    mutate(across(cyl, ~sum(.x)))
#> Error: Problem while computing `..1 = across(cyl, ~sum(.x))`.
#>   i The error occurred in group 0: character(0).
#> Caused by error in `across()`:
#>   Can't subset columns that don't exist.
#>   x Column `cyl` doesn't exist.

^{Created on 2021-12-12 by the reprex package (v2.0.1)}

PS: BTW, in case you are interested, I just made a little GitHub Action that bumps the dev-level package version at each push (link). This would make it easier to tell you the exact version I'm running, as for now it will almost always be x.x.x.9000. I can make a PR if you want.

[hadley]

Making this work would require an amount of effort that I don't think is commensurate with the payoff, so we don't plan to work on this currently.

[GuiFabre]

Hello @hadley and @DanChaltiel

I faced the same issue. The problem to me is that everything() silently bypass the grouping column, while naming it properly doesn't.

# does not work and throw an error
head(mtcars) %>% 
  group_by(cyl) %>% 
  mutate(across(cyl, ~sum(.)))

# does not work but silently bypass cyl (everything is summed, except cyl)
head(mtcars) %>% 
  group_by(cyl) %>% 
  mutate(across(everything(), ~sum(.)))

# does work
head(mtcars) %>% 
  group_by(cyl) %>% 
  mutate(cyl = sum(cyl)))

But to be honest, i understand the decision not to correct it.
In a way or another, I should not make changes into group. Grouped tibble is closer to a list of tibble minus grouping columns which names are groups, rather than a tibble with a specific column.

solution proposed : use rlang synthax to make it work ( !!group_var, any_of(group_var) or quo(cyl) )

group_var <- "cyl"

head(mtcars) %>% 
    group_by(pick(!!group_var)) %>% 
    mutate(!!group_var := sum(!!as.name(group_var)))

Thanks for this thread, I couldn't see where it went wrong :)