Grouping on multiple programmatically specified vars in dplyr 0.6

Question

This is with the new dplyr, to be released Real Soon Now.

The dplyr programming vignette gives an example of calling group_by with the grouping variables specified in an outer function:

my_summarise <- function(df, group_var) {
  df %>%
    group_by(!!group_var) %>%
    summarise(a = mean(a))
}

This works when supplied with a single grouping variable. However, it fails with multiple variables.

Simplified example:

f <- function(x)
{
    group_by(mtcars, !!x)
}

## works
g1 <- "cyl"
f(g1)

## doesn't work
#Error in mutate_impl(.data, dots) : 
#  Column `c("cyl", "gear")` must be length 32 (the number of rows) or one, not 2
g2 <- c("cyl", "gear")
f(g2)

How can I fix this, within the rlang framework?

Ideally I want f's signature to remain the same, ie I specify the grouping variables as a single vector rather than via a ... argument.

Answer 1

There was a pretty similar question: Programming with dplyr using string as input. I just modified the answer a bit to use syms and !!!.

library(rlang)
f <- function(x){
  group_by(mtcars, !!!syms(x))
}

f(c("cyl")) %>% summarise(n())
# A tibble: 3 x 2
    cyl `n()`
  <dbl> <int>
1     4    11
2     6     7
3     8    14

f(c("cyl", "gear")) %>% summarise(n())
# A tibble: 8 x 3
# Groups:   cyl [?]
    cyl  gear `n()`
  <dbl> <dbl> <int>
1     4     3     1
2     4     4     8
3     4     5     2
4     6     3     2
5     6     4     4
6     6     5     1
7     8     3    12
8     8     5     2