In R, how do you evaluate … in the calling function?

Question

If I want to know what is stored in a ... argument within an R function, I can simply convert it to be a list, like so

foo <- function(...)
{
  dots <- list(...)
  print(dots)
}

foo(x = 1, 2, "three")
#$x
#[1] 1
#
#[[2]]
#[1] 2
#
#[[3]]
#[1] "three"

What I can't figure out is how to evaluate ... in the calling function. In this next example I want the contents of baz to return the ... argument to bar.

bar <- function(...)
{
  baz()
}

baz <- function()
{ 
  # What should dots be assigned as?
  # I tried                                           
  # dots <- get("...", envir = parent.frame())
  # and variations of
  # dots <- eval(list(...), envir = parent.frame())
  print(dots)
}

bar(x = 1, 2, "three")

get("...", envir = parent.frame()) returns <...>, which looks promising, but I can't figure out how to extract anything useful from it.

eval(list(...), envir = parent.frame()) throws an error, claiming that ... is used incorrectly.

How can I retrieve the ... from bar?

Answer 1

Figured it out. I needed a combination of eval and substitute. baz should be defined as

baz <- function()
{ 
  dots <- eval(substitute(list(...), env = parent.frame()))
  print(dots)
}

Answer 2

This should work:

bar <- function(...)
{
  baz(...=...)
}

baz <- function(...)
{ 
  print(list(...))
}

bar(x = 1, 2, "three")

Just assign it in the subfunction.

Alternatively, you can assign the ellipsis as a list in the parent function:

bar <- function(...)
{
  bar.x <- list(...)
  baz()
}

baz <- function()
{ 
  dots <- get("bar.x", envir = parent.frame())
  print(dots)
}

bar(x = 1, 2, "three")

This is the same idea, but I wouldn't suggest it because you're overwriting the ellipsis:

bar <- function(...)
{
  ... <- list(...)
  baz()
}

baz <- function()
{ 
  dots <- get("...", envir = parent.frame())
  print(dots)
}

bar(x = 1, 2, "three")

Answer 3

In a word: don't. Trying to redefine R's scoping rules is only likely to end up in heartache and pain.