R missing() with variable names

Question

In R 3.0.2 the missing() function can tell us whether or not a formal parameter is missing.

How can one avoid hardcoding the variable name passed into missing? e.g. in

demoargs <- function(a=3, b=2, d) {
    f <- formals(demoargs)  # Capture formal arguments
    formalNames <- names(f) # get variable names: a, b, d
    ...   
}

I would like to be able to check for missing formals without doing this in a hardcoded manner, e.g.:

missing(formalNames[1])  # returns invalid use of missing!

as opposed to missing(d) for the purpose of iterating over a large number of optional arguments that are handled in a limited number of manners. I had hoped that get or as.name my put me on the right track but this does not seem to be the case.

Alternatively, I suspect that I could do this with the vararg arguments (...), but it would be nice for the caller to be able to inspect the acceptable optional arguments by examining the function declaration.

Thanks, Marie

Answer 1

You probably first tried something like missing(as.name(formalNames[1])) or missing(formalNames[1]) and found that they do not work.

The reason they don't is that missing() is one of those odd functions -- library() and debug() are a couple of others -- that will accept as an argument either a name or a character representation of a name. That's 'nice' in the sense that missing(a) and missing("a") will both check whether the function call included a supplied argument a; it's not so nice when you do missing(formalNames[1]) and it goes off to look for a non-existent argument named formalNames[1].

The solution is to use do.call(), which evaluates the elements of its second argument before passing them on to the function given in its first argument. Here's what you might do:

demoargs <- function(a=3, b=2, d) {
    formalNames <- names(formals()) # get variable names: a, b, d
    do.call(missing, list(formalNames[1]))
}

## Try it out
demoargs(a=42)
# [1] FALSE
demoargs()
# [1] TRUE