Best practice question - elipsis or array parameters?

Question

When I design APIs I have the choice between something like
    public static void foo(String[] params)
        // do stuff
    }or
    public static void foo(String... params)
        // do stuff
    }If I have both declarations I get the following error (which I can understand):
"cannot declare both foo(java.lang.String...) and foo(java.lang.String[]) in blablabla.Main
public static void foo(String... params)"
So I can use one or the other. The question is: What is "better"?
The elipsis version (with ...) looks nicer because I don't have to build an array prior to calling foo if my params do come from separate sources and I can pass them both separately.

Answer 1

But isn't this the same as a call to foo(null) if I would have used the array version?Depends on how you implemented your method. Example:
public static void foo1(String[] params) {
  for(String param: params) {
    System.out.println(param.length());
public static void foo2(String... params) {
  for(String param: params) {
    System.out.println(param.length());
public static void main(String[] args) {
  foo1(null);
  foo2();
}The call foo1(null) would fail for obvious reasons, whereas foo2() would be valid.
Lesson: an empty array is different from a null array.
However, is there any other consideration I should do - performance for instance? Forget about performance: ellipsis is just a convenient way to overload a method for the no-arg case, so use it if you have a no arg-case, else use the array version.