Question

When I look at R functions I often find the following structure:

f <- function(exp=T) {
  if (exp)
    a <- 1
  else
    a <- 2
}
f()
f(F)

This will run without error. But executing the inner function code throws an error as R probably assumes that the statement is finished after the first assignment a <- 1 and cannot handle the following else.

exp=T
if (exp)
  a <- 1
else
  a <- 2

Now, this makes sense to me, but I still would like to understand why the behaviour of the executed code differs when executed inside or outside a function.

Answer 1

It’s a consequence of using an interactive shell (REPL) to run scripts:

After the first branch the shell has seen a complete statement so it assumes that you’re done typing. Unfortunately, R uses the same shell for interpreting scripts even if they are not typed in interactively – so even when you save the if statement to a file and source it (or pipe it into R) you will get the error on the else branch.

But the following will work just fine:

if (exp) a <- 1 else a <- 2

Here, the interpreter swallows the line and executes it.

In your function one would assume that the same applies – and it does! However, the function itself starts with an open brace in your case, so R has to read until it finds the matching closing brace. By contrast, take this function declaration:

f <- function (exp)
    if (exp)
        a <- 1
    else
        a <- 2

In R you can define functions without braces around the body. But the above code will fail for the same reason that the standalone if without braces fails. By contrast, if I had written the if on a single line this code would once again work.

Incidentally, your function uses an assignment to a variable that isn’t used. You can (should) do the following instead:

f <- function (exp) {
    if (exp)
        1
    else
        2
}

… and the same when using if inside the shell:

a <- if (exp) 1 else 2

because in R, if is an expression which returns a value.