Subj : Re: Equivalent of JS_InitClass for global object?
To   : netscape.public.mozilla.jseng
From : Brendan Eich <brendan@meer.net>
Date : Mon Aug 25 2003 02:00 pm

Brendan Eich wrote:

> John J Lee wrote:
>
>> I'll take your word for that, but I don't understand why (I *do* still
>> have window objects in Python -- it's just that rather than bind one
>> directly as the global object, I just have a function bind_window that
>> fakes a JS global object by grabbing the individual attributes and 
>> methods
>> from a window object and binding them to my simple JS global object).
>>
>
> Windows are global objects in the DOM level 0, so a variable (var x = 
> 42, e.g.) or function declaration (function f() {return 42;} at 
> top-level) finds


I meant to write "binds" here, not "finds" (although for the variable 
case, per ECMA-262, if a property named x already existed when the 
execution context for 'var x = 42' was entered, that pre-existing 
property would be used; it would not be blown away and replaced by a new 
property -- and so, if it were read-only, you'd get a runtime error 
trying to set it to 42).

/be

> a property of the global object, which can be fetched via window.x or 
> window.f, or if in another window accessible according to security 
> policy and frame/framset or window.open call prerequisites, 
> otherWindow.x or otherWindow.f.
>
> Plus, you can count on window object identity invariance: 
> otherWindow.myWindow = window; <load a new document in window>; 
> assert(otherWindow.myWindow == window).
>
> Does your window object not being the global object magically satisfy 
> these constraints?
>
>> What's the best way to find out about this kind of thing?  ATM, I don't
>> know, for example, whether all these frames / windows have their own
>> scopes / global objects / JSContexts -- your answer seems to hint at
>> something like this.  Am I stuck with either attempting to read the
>> Mozilla source, or guessing by looking at piles of real-world JS code?
>>
>
> Other browsers have managed to implement the DOM level 0, even though 
> the w3c never stepped up and standardized it.  Perhaps the question of 
> specifying it enough for you to implement it without reading Mozilla 
> sources should be turned over to the w3c.  Or to the 
> news://news.mozilla.org/netscape.public.mozilla.dom newsgroup, at least.
>
> /be
>

.