Search code examples
rdplyrtidyversetidyeval

Passing multiple arguments with tidyeval - different methods?

I'm trying to understand tidyeval and came across a tutorial from Lionel Henry. I've also read the vignette from dplyr. They describe different methods of passing multiple arguments, e.g. when using group_by.

What's the difference between the two examples below? In simple terms(if possible): Which one should I use and why? 

library(dplyr)

mean_by <- function(data, var, ...) {
  var <- enquo(var)

  data %>% 
    group_by(...) %>% 
    summarise(avg = mean(!! var))
}

mean_by2 <- function(data, var, ...) {
  var <- enquo(var)
  group <- quos(...)

  data %>% 
    group_by(!!! group) %>% 
    summarise(avg = mean(!! var))
}

all_equal(mean_by(starwars, height, species, eye_color),
          mean_by2(starwars, height, species, eye_color))
#> [1] TRUE


Created on 2018-10-05 by the [reprex
package](http://reprex.tidyverse.org) (v0.2.0).
Solution

  • They are functionally equivalent. If you're not modifying the arguments in the dots (changing their names, wrapping a function around the arguments, ...), it's better to pass the ... directly as it's more concise and readable.

    This section of the tidy eval bookdown explores these issues: https://tidyeval.tidyverse.org/multiple.html