Subj : Re: Array and [ ]
To   : =?ISO-8859-1?Q?Georg_Maa=DF?=
From : Brendan Eich
Date : Sun Jan 23 2005 09:29 am

Georg Maaß wrote:
> Brendan Eich wrote:
> 
>>> The production ArrayLiteral : [ Elision opt ] is evaluated as follows:
>>> 1. Create a new array as if by the expression "new Array()"
>>
>>
>> The spec is unambiguous here.  Elsewhere, it talks about "the original 
>> value of the Object.prototype property", or similar such words.  This 
>> should be fixedd at some point, but I haven't managed to get the ECMA 
>> TG1 group as a whole to go through the 
>> http://www.mozilla.org/js/language/E262-3-errata.html yet, let alone 
>> consider this question.
> 
> 
> There is no entry about 11.1.4.


I know -- I said "let alone" meaning that the current topic is beyond 
the known errata.  We still haven't agreed on the right spec.


>> Another way to fix things is to make Array readonly and permanent in 
>> the global object (likewise all standard class constructors and other 
>> predefined global properties).
> 
> 
> If it is not specified as read only then making it read only means 
> implementing a bug. There is no need to make it impossible to users to 
> customize their Array constructor.


I was not talking about implementing a bug, I was talking about another 
way to fix the spec.  Personally I prefer the current SpiderMonkey way, 
and if you want to be able to change Array (as shown in your previous 
post, to lowercase string elements added to instances, e.g. -- although 
a subclass of Array might be more appropriate in a real application), 
more power to you.

But if you want Array to be mutable, then I think you should want [] to 
be short-hand for new Array.

/be

.