Creating a function in R with variable number of *named* arguments

Question

In R, one can create a function with a variable number of arguments using ... , as illustrated here .

However, I would like to have a function with a variable number of named (formal) arguments. As a minimal example, something along the lines of

f_nargs <- function(n) {

  # process argument n
  # ...
  f(x1, ..., xn) # What should this be ?
}

How can I achieve this?

---

The overall goal is to use a function(al) in a package. That function(al) requires as input a function with 3 formal arguments. While I could write the function specially for n=3, I am curious if this can be done for a general n.

---

Edit : Here is what I have in mind:

My current function is f(x) , where x is a vector of length 3.

I would like to convert this into f(x1, x2, x3) before passing it to the functional in the package. This functional checks if length(formals(f))==3 .

Currently, I pass f into

split_args3 <- function(f) {
  force(f)
  function(x1, x2, x3, ...) {
    f(c(x1, x2, x3), ...)
  }
}

This works, but only for a specific n. I then tried using

split_args <- function(f, d) {
  force(f)
  force(d)
  function(...) {
    dots <- list(...)
    arg1 <- do.call(c, dots[1:d])
    do.call(f, append(list(arg1), dots[-(1:d)]) )
  }
}

but it fails the check in the functional.

Answer 1

Alter the list of arguments how you like, then pass it into do.call :

split_args <- function(f, d) {
    force(f)
    force(d)
    function(...) {
        dots <- list(...)
        dots <- c(list(unlist(dots[seq(d)])), dots[-seq(d)])
        do.call(f, dots)
    }
}

mean3 <- split_args(mean, 3)
mean3(2, NA, 3, na.rm = TRUE)
#> [1] 2.5