disctint() doesn't respect the order of variables provided

[fmichonneau]

The output of distinct() doesn't follow the order of the columns provided by the user but uses the order of the original data:

> starwars %>% distinct(hair_color, skin_color)
# A tibble: 50 x 2
      hair_color  skin_color
           <chr>       <chr>
 1         blond        fair
 2          <NA>        gold
 3          <NA> white, blue
 4          none       white
 5         brown       light
 6   brown, grey       light
 7          <NA>  white, red
 8         black       light
 9 auburn, white        fair
10  auburn, grey        fair
# ... with 40 more rows
> starwars %>% distinct(skin_color, hair_color)
# A tibble: 50 x 2
      hair_color  skin_color
           <chr>       <chr>
 1         blond        fair
 2          <NA>        gold
 3          <NA> white, blue
 4          none       white
 5         brown       light
 6   brown, grey       light
 7          <NA>  white, red
 8         black       light
 9 auburn, white        fair
10  auburn, grey        fair
# ... with 40 more rows

[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/