add_count() and add_tally() should error if output column name is already used

[ghost]

I'm wondering if add_count() and add_tally() should throw an error in case that the output column name (default "n") is already used in a pre-existing column. Otherwise it ends up doing some odd behaviour. For instance:

library(dplyr)
#> 
#> Attaching package: 'dplyr'
#> The following objects are masked from 'package:stats':
#> 
#>     filter, lag
#> The following objects are masked from 'package:base':
#> 
#>     intersect, setdiff, setequal, union

df <- tibble(a = rep(LETTERS, 2), n = 1:52)

df
#> # A tibble: 52 x 2
#>    a         n
#>    <chr> <int>
#>  1 A         1
#>  2 B         2
#>  3 C         3
#>  4 D         4
#>  5 E         5
#>  6 F         6
#>  7 G         7
#>  8 H         8
#>  9 I         9
#> 10 J        10
#> # ... with 42 more rows

add_tally(df)
#> Using `n` as weighting variable
#> # A tibble: 52 x 2
#>    a         n
#>    <chr> <int>
#>  1 A      1378
#>  2 B      1378
#>  3 C      1378
#>  4 D      1378
#>  5 E      1378
#>  6 F      1378
#>  7 G      1378
#>  8 H      1378
#>  9 I      1378
#> 10 J      1378
#> # ... with 42 more rows

Do we want n to be used as weighting variable? This behaviour seems unexpected to me.

add_count(df, a)
#> # A tibble: 52 x 2
#>    a         n
#>    <chr> <int>
#>  1 A         2
#>  2 B         2
#>  3 C         2
#>  4 D         2
#>  5 E         2
#>  6 F         2
#>  7 G         2
#>  8 H         2
#>  9 I         2
#> 10 J         2
#> # ... with 42 more rows

In this case, add_count() silently replaces the pre-existing column "n" with its output, which is likely not the user's intent.

EDIT: I see that the behaviour with add_count is already discussed here. Fair enough on that front.

[nschmucker]

It seems the behavior of add_tally() changed in version 0.8.0.1. Previously, in 0.7.6...

> library(dplyr)

> mtcars %>% add_tally(vs) %>% add_tally(am)
# A tibble: 32 x 13
     mpg   cyl  disp    hp  drat    wt  qsec    vs    am  gear  carb     n    nn
   <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
 1  21       6  160    110  3.9   2.62  16.5     0     1     4     4    14    13
 2  21       6  160    110  3.9   2.88  17.0     0     1     4     4    14    13
 3  22.8     4  108     93  3.85  2.32  18.6     1     1     4     1    14    13
 4  21.4     6  258    110  3.08  3.22  19.4     1     0     3     1    14    13
 5  18.7     8  360    175  3.15  3.44  17.0     0     0     3     2    14    13
 6  18.1     6  225    105  2.76  3.46  20.2     1     0     3     1    14    13
 7  14.3     8  360    245  3.21  3.57  15.8     0     0     3     4    14    13
 8  24.4     4  147.    62  3.69  3.19  20       1     0     4     2    14    13
 9  22.8     4  141.    95  3.92  3.15  22.9     1     0     4     2    14    13
10  19.2     6  168.   123  3.92  3.44  18.3     1     0     4     4    14    13
# ... with 22 more rows

Now, though, under 0.8.0.1, it overwrites the n column, rather than automatically adding a nn column.

> mtcars %>% add_tally(vs) %>% add_tally(am)
# A tibble: 32 x 12
     mpg   cyl  disp    hp  drat    wt  qsec    vs    am  gear  carb     n
   <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
 1  21       6  160    110  3.9   2.62  16.5     0     1     4     4    13
 2  21       6  160    110  3.9   2.88  17.0     0     1     4     4    13
 3  22.8     4  108     93  3.85  2.32  18.6     1     1     4     1    13
 4  21.4     6  258    110  3.08  3.22  19.4     1     0     3     1    13
 5  18.7     8  360    175  3.15  3.44  17.0     0     0     3     2    13
 6  18.1     6  225    105  2.76  3.46  20.2     1     0     3     1    13
 7  14.3     8  360    245  3.21  3.57  15.8     0     0     3     4    13
 8  24.4     4  147.    62  3.69  3.19  20       1     0     4     2    13
 9  22.8     4  141.    95  3.92  3.15  22.9     1     0     4     2    13
10  19.2     6  168.   123  3.92  3.44  18.3     1     0     4     4    13
# ... with 22 more rows

I like the new option to specify a column name (!) but it seems that existing code that relied on the nn column name may run into issues. This seems somewhat related to the documentation change noted in #4124

[nteetor]

Ran into the problem @nschmucker brings up. I went looking for a mention in NEWS, but found nothing.

[hadley]

Now:

library(dplyr, warn.conflicts = FALSE)
df <- tibble(g = c(1, 1, 2), n = c(1, 1, 10))

df %>% count(g)
#> Using `n` as weighting variable
#> Error: Column 'n' is already present in output
#>  * Use `name = "new_name"` to pick a new name
df %>% add_count(g)
#> Using `n` as weighting variable
#> Error: Column 'n' is already present in output
#>  * Use `name = "new_name"` to pick a new name

# Create new variable
df %>% count(g, name = "nn")
#> Using `n` as weighting variable
#> # A tibble: 2 x 2
#>       g    nn
#>   <dbl> <dbl>
#> 1     1     2
#> 2     2    10
df %>% add_count(g, name = "nn")
#> Using `n` as weighting variable
#> # A tibble: 3 x 3
#>       g     n    nn
#>   <dbl> <dbl> <dbl>
#> 1     1     1     2
#> 2     1     1     2
#> 3     2    10    10

# Overide existing variable
df %>% count(g, name = "n")
#> Using `n` as weighting variable
#> # A tibble: 2 x 2
#>       g     n
#>   <dbl> <dbl>
#> 1     1     2
#> 2     2    10
df %>% add_count(g, name = "n")
#> Using `n` as weighting variable
#> # A tibble: 3 x 2
#>       g     n
#>   <dbl> <dbl>
#> 1     1     2
#> 2     1     2
#> 3     2    10

^{Created on 2019-12-31 by the reprex package (v0.3.0)}

(and similarly for tally())