What's going on when mutate RHS is a data frame?

[doorisajar]

Recently I came across a use case where a colleague wanted to duplicate a column via mutate, but refer to the target column programmatically. In the course of doing this, I checked to see what would happen if I used [ rather than [[, and found that not only did mutate not fail, but I got a strange result:

library(dplyr)

my_col <- "base_col"

df <- data.frame("a" = c(1, 2),
                 "b" = c(2, 3),
                 "base_col" = c(3, 4))

# this works fine
df %>% mutate(new_col = .[[my_col]])
#>   a b base_col new_col
#> 1 1 2        3       3
#> 2 2 3        4       4

# what's going on here?
df %>% mutate(new_col = .[my_col])
#>   a b base_col new_col. new_col.
#> 1 1 2        3        3        3
#> 2 2 3        4        4        4

mutate is receiving a data.frame here:

> df[my_col]
  base_col
1        3
2        4

...but it isn't obvious why the column gets added twice or where the periods are coming from.

I think the right behaviour here is probably to fail, unless I'm missing something?

[krlmlr]

Agree that we should throw an error here, but maybe just throw a warning first.

It does create a new column new_col that contains a data frame without column names. Subsequent dplyr operations might fail on that structure.

Look at str(df %>% ...) or dput(df %>% ...) to see the internal structure of the result.

[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/