Why are anonymous function expressions and named f

问题:

I'm looking at section 13 or the ECMAScript specification (v. 5). An anonymous function expression is initialized as follows:

Return the result of creating a new Function object as specified in 13.2 with parameters specified by FormalParameterListopt and body specified by FunctionBody. Pass in the LexicalEnvironment of the running execution context as the Scope. Pass in true as the Strict flag if the FunctionExpression is contained in strict code or if its FunctionBody is strict code.

this logic is very similar to how a function declaration is initialized. However, notice how different initialization of a named funciton expression is.

1. Let funcEnv be the result of calling NewDeclarativeEnvironment passing the running execution context’s Lexical Environment as the argument
2. Let envRec be funcEnv’s environment record.
3. Call the CreateImmutableBinding concrete method of envRec passing the String value of Identifier as the argument.
4. Let closure be the result of creating a new Function object as specified in 13.2 with parameters specified by FormalParameterListopt and body specified by FunctionBody. Pass in funcEnv as the Scope. Pass in true as the Strict flag if the FunctionExpression is contained in strict code or if its FunctionBody is strict code.
5. Call the InitializeImmutableBinding concrete method of envRec passing the String value of Identifier and closure as the arguments.
6. Return closure.

I know one of the big differences between named/anonymous function expressions is that named function expressions can be called recursively from within the function, but that's all I can think of. Why is the setup so different and why does it need to do those extra steps?

回答1:

The reason for all that "dancing" is simple.

The identifier of named function expression needs to be made available within function scope but not outside.

typeof f; // undefined

(function f() {
  typeof f; // function
})();

How do you make f available within function?

You can't create binding in the outer Lexical Environment since f shouldn't be available outside. And you can't create binding in the inner Variable Environment since... it's not yet created; the function is not yet executed at the moment of instantiation, and so 10.4.3 (Entering Function Code) step with its NewDeclarativeEnvironment has never happened.

So the way this is done is by creating an intermediate lexical environment that "inherits" directly from current one, and which is then passed as [[Scope]] into the newly created function.

You can see this clearly if we break steps in 13 into pseudo code:

// create new binding layer
funcEnv = NewDeclarativeEnvironment(current Lexical Environment)

envRec = funcEnv
// give it function's identifier
envRec.CreateImmutableBinding(Identifier)

// create function with this intermediate binding layer
closure = CreateNewFunction(funcEnv)

// assign newly created function to an identifier within this intermediate binding layer
envRec.InitializeImmutableBinding(Identifier, closure)

So the lexical environment within f (when resolving identifier, for example) now looks like this:

(function f(){

  [global environment] <- [f: function(){}] <- [Current Variable Environment]

})();

With anonymous function it would look like this:

(function() {

  [global environment] <- [Current Variable Environment]

})();

回答2:

The core difference between the two deals with scoping (although, if nothing else, it's curious to see how much is actually involved in doing so ;) - and you're already correct in pointing out the main difference between named/anonymous function expresssions being the ease of calling the named ones recursively.

Why do I say ease? Well, nothing actually prevents you from calling an anonymous function recursively but it's just plain not pretty:

//silly factorial, 5!
(function(n) {
  if (n<=1) return 1;
  return (n*arguments.callee(n-1)); //arguments.callee is so 1990s!
})(5);

In fact, that's exactly what MDN says in describing named function expressions!

If you want to refer to the current function inside the function body, you need to create a named function expression. This name is then local only to the function body (scope). This also avoids using the non-standard arguments.callee property.