Subj : Re: Error in decompiling
To   : netscape.public.mozilla.jseng
From : Martin Honnen <Martin.Honnen@t-online.de>
Date : Sun Jan 18 2004 07:29 pm



Adrian Klein wrote:

> Assume a function defined as:
> 
>     function do_something() {
>       var obj = new Object();
> 
>       obj['name']        = 'Just a name';
>       obj['if']        = false;
>           obj['some text'] = 'correct';
>         }
> 
> do_something.toString(), do_something.String() or uneval(do_something) 
> will be resolved to something like this (with varying formating):
> 
>     function do_something() {
>       var obj = new Object();
> 
>       obj.name        = 'Just a name';
>       obj.if        = false;
>           obj['some text'] = 'correct';
>         }
> 
> I think this seems like a bug in the decompiling part of the 
> Spidermonkey Engine, because it does not check if the name of a object 
> property is a reserved keyword, so that it should not be refered to with 
> the 'object.property' syntax.

ECMAScript edition 3 says about the toString method of function objects:

15.3.4.2 Function.prototype.toString ( )
An implementation-dependent representation of the function is returned. 
This representation has the
syntax of a FunctionDeclaration.

so one can indeed call the above a bug as the result of toString() 
doesn't yield a representation fitting the production for 
FunctionDeclaration.

When I try Rhino with a similar example it gets it right:

Rhino 1.5 release 4.1 2003 04 21
js> function f () { var obj = new Object(); obj['if'] = false; }
js> f.toString()

function f() {
     var obj = new Object();
     obj["if"] = false;
}


However if I use an object literal then Rhino gets that wrong

Rhino 1.5 release 4.1 2003 04 21
js> function f () { var obj = {'if': false}; }
js> f.toString()

function f() {
     var obj = {if:false};
}


You could file a bug on bugzilla (or wait what Brendan thinks about it).

-- 

	Martin Honnen
	http://JavaScript.FAQTs.com/

.