Subj : Re: Weird nested function behavior in IE
To   : Ryan Breen
From : Igor Bukanov
Date : Tue Nov 02 2004 11:42 pm

Ryan Breen wrote:
> The following:
> 
> if(true)
> {
>     outer.prototype.inner=hello_world;
>     function outer()
>     {
>        this.inner();
>     }
> 
>     function hello_world()
>     {
>        alert("HI");
>     }
> 
>     var hold_me =new outer();
> }
> 
> works in IE but fails in Firefox or Rhino.  In Rhino, the error is
> that 'outer' is evaluated as a FUNCTION_EXPRESSION_STATEMENT rather
> than a FUNCTION_STATEMENT but fn.itsClosure is null.  Has anyone seen
> anything like this before?

Note that this code does not confirm to ECMAScript standard since 
ECMAScript allows function statements only at the top level of script or 
function.

Now MS JScript and SpiderMonkey/Rhino extends that allows to define 
function statements anywhere were a JS statement is allowed, but the 
semantics is different:

In JScript all such function statements are treated in the same way and 
effectively evaluated at the start of script/function so you can refer 
the functions by names right away.

In SM/Rhino such functions are defined only when the corresponding code 
reaches the point of the "extended function statement" declaration which 
allows to have a conditional compilation.

For example given:

if (condition) {
     function test1() {
     }
}

function test2() {
}


then JScript would always define test1 even if condition is false while 
SM/Rhino would define it only when condition is true.

To make you example to work move hello_world()/outer to the beginning of 
the if block.

Regards, Igor

.