Correct way to specifiy optional arguments in R functions?

Question

I am interested in what is the "correct" way to write functions with optional arguments in R.Over time, I stumbled upon a few pieces of code that take a different route here, and I couldn't find a proper (official) position on this topic.Up until now, I have written optional arguments like this: fooBar <- function(x,y=NULL){ if(!is.null(y)) x <- x+y return(x) } fooBar(3) # 3 fooBar(3,1.5) # 4.5The function simply returns its argument if only `x` is supplied. It uses a default `NULL` value for the second argument and if that argument happens to be not `NULL`, then the function adds the two numbers.Alternatively, one could write the function like this (where the second argument needs to be specified by name, but one could also `unlist(z)` or define `z <- sum(...)` instead): fooBar <- function(x,...){ z <- list(...) if(!is.null(z$y)) x <- x+z$y return(x) } fooBar(3) # 3 fooBar(3,y=1.5) # 4.5Personally I prefer the first version. However, I can see good and bad with both. The first version is a little less prone to error, but the second one could be used to incorporate an arbitrary number of optionals.Is there a "correct" way to specify optional arguments in R? So far, I have settled on the first approach, but both can occasionally feel a bit "hacky".

Answer 1

While either will work, I prefer the former. "..." usually indicates a set of "passthrough" variables that are meaningful to functions invoked by your function. If, instead, this indicates a suite of optional variables for possible use within the function itself, you have simplified the function definition, but likely made it more difficult for other users to decipher. I go this route only when I consider the optional variables to be "style" related (e.g., changing some stylistics aspect of a plot).

That's not set in stone, of course. Nothing in R is set in stone.