Best practice, or doesn't really matter?

[anakadote]

Shifting from procedural to OOP, so say I'm going to delete a user, I used to have something like:

if(!empty($_GET['d_uid'])){
  $uid = mysql_real_escape_string($_GET['d_uid']);
  if(deleteUser($uid)){
    // success
  } else {
    // fail
  }
}

But now, since I have written a class to hold all the user functions (class name User), is the best was to accomplish this to instantiate the user class in order to access a function for deleting the user?

Example:

if(!empty($_GET['d_uid'])){
  $uid = mysql_real_escape_string($_GET['d_uid']);
  $user = new User($uid);
  if($user->deleteUser()){
    // success
  } else {
    // fail
  }
}

or perhaps use a scope resolution operator?:

if(!empty($_GET['d_uid'])){
  $uid = mysql_real_escape_string($_GET['d_uid']);
  if(User::deleteUser($uid)){
    // success
  } else {
    // fail
  }
} 

[NogDog]

It may be better to look at this from a top down view: what is the object to which this method will belong, and what does that object model? If the object is only going to be a set of utility methods related to the users table, it could conceivably just be a static method that you could then access directly as you suggest with the "::" operator. On the other hand, if you visualize an object which represents a single user with methods applicable to dealing with that single user, then it would more likely be something where you instantiate the object, specify which user that object represents, and then the delete method would not even have any argument (as the user ID would already have been specified when populating the object, perhaps via the constructor).

[anakadote]

Great, that makes perfect sense. Thanks!