Prev: Setter/Getter in module pattern?
Next: FAQ Topic - What is JScript? (2010-07-16)
From: kangax on 16 Jul 2010 19:44
On 7/16/10 7:31 PM, Thomas 'PointedEars' Lahn wrote:
> kangax wrote:
>
>> Thomas 'PointedEars' Lahn wrote:
>>> RobG wrote:
>> [...]
>>> ECMAScript Edition 5 specifies the Object.defineProperty() and
>>> Object.defineProperties() methods to define a non-enumerable property;
>>> they are implemented in Google V8 since version 2.1 (Chrome 5.0.342) and
>>> Apple JavaScriptCore since version 533.16 (Safari 4.0.4).
>> ^^^^^
>>
>> That should be Safari 5, not 4.0.4 (build number is right, though).
>>
>> Also see<http://kangax.github.com/es5-compat-table/>
>
> No. As a matter of fact, "Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US)
> AppleWebKit/533.16 (KHTML, like Gecko) Version/4.0.4 Safari/531.21.10"
> supports both Object.defineProperty() and Object.defineProperties().

Interesting...

My Safari 4.0.5 [1] on Mac OS X has neither `Object.defineProperty` nor
`Object.defineProperties`; Safari 5 [2] has both.

[1] Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10_6_4; en-us)
AppleWebKit/531.22.7 (KHTML, like Gecko) Version/4.0.5 Safari/531.22.7

[2] Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10_6_4; en-us)
AppleWebKit/533.16 (KHTML, like Gecko) Version/5.0 Safari/533.16

I don't understand how your 4.0.4 ended up with 533.x webkit build.

Can anyone else confirm?

--
kangax
From: Thomas 'PointedEars' Lahn on 16 Jul 2010 20:31
kangax wrote:

> Thomas 'PointedEars' Lahn wrote:
>> kangax wrote:
>>> Thomas 'PointedEars' Lahn wrote:
>>>> RobG wrote:
>>> [...]
>>>> ECMAScript Edition 5 specifies the Object.defineProperty() and
>>>> Object.defineProperties() methods to define a non-enumerable property;
>>>> they are implemented in Google V8 since version 2.1 (Chrome 5.0.342)
>>>> and Apple JavaScriptCore since version 533.16 (Safari 4.0.4).
>>> ^^^^^
>>>
>>> That should be Safari 5, not 4.0.4 (build number is right, though).
>>>
>>> Also see<http://kangax.github.com/es5-compat-table/>
>>
>> No. As a matter of fact, "Mozilla/5.0 (Windows; U; Windows NT 5.1;
>> en-US) AppleWebKit/533.16 (KHTML, like Gecko) Version/4.0.4
>> Safari/531.21.10" supports both Object.defineProperty() and
>> Object.defineProperties().
>
> Interesting...
>
> My Safari 4.0.5 [1] on Mac OS X has neither `Object.defineProperty` nor
> `Object.defineProperties`; Safari 5 [2] has both.
>
> [1] Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10_6_4; en-us)
> AppleWebKit/531.22.7 (KHTML, like Gecko) Version/4.0.5 Safari/531.22.7
>
> [2] Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10_6_4; en-us)
> AppleWebKit/533.16 (KHTML, like Gecko) Version/5.0 Safari/533.16
>
> I don't understand how your 4.0.4 ended up with 533.x webkit build.

JFTR: I had the older version before, but thought it was an error to be
corrected. (So I did. D'oh.)

Now that you pointed out the version mismatch too, I came to suspect that it
was because I had installed Safari 5.0 in the same Wine subtree. Since I
suddenly had problems running any Safari browser tonight (which might have
to do with a winecfg on the wrong subtree), I have just reinstalled version
4.0.4 again into a clean subtree, and you are right:

"Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US) AppleWebKit/531.21.8
(KHTML, like Gecko) Version/4.0.4 Safari/531.21.10" does not support
Object.defineProperty() or Object.defineProperties(). (See the tests
below.)

Thank you. (Note to self: Always install a Safari version into its own Wine
subtree.)

It would appear that I have to double-check all JavaScriptCore test results
for the ECMAScript Support Matrix¹ so far. BTW, those are tests for
functionality, and will be available with the next revision of the Matrix.
As a preview, for those two methods they currently are (PHP code cleaned,
pretty-printed):

var o = new Object(),
b = isMethod(Object, 'defineProperty')
&& Object.defineProperty(o, 'a', {
value: {b: 'c'},
writable: false,
configurable: false,
enumerable: false
})
&& (typeof o.a == 'object') && o.a
&& (o.a.b == 'c')
&& (o.a = 42) && (o.a != 42);
delete o.a;
b = b && (typeof o.a != 'undefined');
if (b)
{
var found = false;
for (var p in o)
{
if (p == 'a')
{
found = true;
break;
}
}
}
b && !found;

and

var o = new Object(),
b = isMethod(Object, 'defineProperties')
&& Object.defineProperties(o, {
a: {
value: {b: 'c'},
writable: false,
configurable: false,
enumerable: false
},
b: {
value: {c: 'd'},
writable: false,
configurable: false,
enumerable: false
}
})
&& (typeof o.a == 'object') && o.a
&& (o.a.b == 'c')
&& (o.a = 42) && (o.a != 42)
&& (typeof o.b == 'object') && o.b
&& (o.b.c == 'd')
&& (o.b = 42) && (o.b != 42);
delete o.a;
delete o.b;
b = b && (typeof o.a != 'undefined') && (typeof o.b != 'undefined');
if (b)
{
var found = false;
for (var p in o)
{
if (p == 'a' || p == 'b')
{
found = true;
break;
}
}
}
b && !found;

The result of either program determines if the corresponding feature is
considered to be supported by an implementation. A true-value indicates
that it is supported, a false-value that it is not.

AFAICS, the tests are only incomplete in that they do not test that the
attributes of a property that does not have the [[Configurable]] attribute
set must not be possible to redefine with Object.defineProperty() (except of
its value).

Suggestions welcome.


PointedEars
___________
¹ <http://PointedEars.de/es-matrix>
--
var bugRiddenCrashPronePieceOfJunk = (
navigator.userAgent.indexOf('MSIE 5') != -1
&& navigator.userAgent.indexOf('Mac') != -1
) // Plone, register_function.js:16
From: kangax on 18 Jul 2010 08:55
On 7/16/10 8:31 PM, Thomas 'PointedEars' Lahn wrote:
> kangax wrote:
[...]
>> My Safari 4.0.5 [1] on Mac OS X has neither `Object.defineProperty` nor
>> `Object.defineProperties`; Safari 5 [2] has both.
>>
>> [1] Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10_6_4; en-us)
>> AppleWebKit/531.22.7 (KHTML, like Gecko) Version/4.0.5 Safari/531.22.7
>>
>> [2] Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10_6_4; en-us)
>> AppleWebKit/533.16 (KHTML, like Gecko) Version/5.0 Safari/533.16
>>
>> I don't understand how your 4.0.4 ended up with 533.x webkit build.
>
> JFTR: I had the older version before, but thought it was an error to be
> corrected. (So I did. D'oh.)
>
> Now that you pointed out the version mismatch too, I came to suspect that it
> was because I had installed Safari 5.0 in the same Wine subtree. Since I
> suddenly had problems running any Safari browser tonight (which might have
> to do with a winecfg on the wrong subtree), I have just reinstalled version
> 4.0.4 again into a clean subtree, and you are right:
>
> "Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US) AppleWebKit/531.21.8
> (KHTML, like Gecko) Version/4.0.4 Safari/531.21.10" does not support
> Object.defineProperty() or Object.defineProperties(). (See the tests
> below.)
>
> Thank you. (Note to self: Always install a Safari version into its own Wine
> subtree.)

No problem. You got me worried there for a bit.

>
> It would appear that I have to double-check all JavaScriptCore test results
> for the ECMAScript Support Matrix¹ so far. BTW, those are tests for
> functionality, and will be available with the next revision of the Matrix.
> As a preview, for those two methods they currently are (PHP code cleaned,
> pretty-printed):
>
> var o = new Object(),
> b = isMethod(Object, 'defineProperty')
> && Object.defineProperty(o, 'a', {
> value: {b: 'c'},
> writable: false,
^^^^^
> configurable: false,
^^^^^
> enumerable: false
^^^^^
> })

Technically, those are default values, so explicitly setting them is
redundant (although adds to clarity).

> && (typeof o.a == 'object')&& o.a
> && (o.a.b == 'c')
> && (o.a = 42)&& (o.a != 42);
> delete o.a;
> b = b&& (typeof o.a != 'undefined');
> if (b)
> {
> var found = false;
> for (var p in o)
> {
> if (p == 'a')

Maybe add `hasOwnProperty` check too?

> {
> found = true;
> break;
> }
> }
> }
> b&& !found;
>

[snip defineProperties test]

>
> The result of either program determines if the corresponding feature is
> considered to be supported by an implementation. A true-value indicates
> that it is supported, a false-value that it is not.
>
> AFAICS, the tests are only incomplete in that they do not test that the
> attributes of a property that does not have the [[Configurable]] attribute
> set must not be possible to redefine with Object.defineProperty() (except of
> its value).

Well if we're talking about full conformance, then there are also no
tests for `Object.defineProperty` throwing TypeError when first argument
is not an object; no test for getters/setters (i.e. "get" and "set"
properties in object corresponding to property descriptor); no test for
ToPropertyDescriptor which is invoked during `defineProperty` and is,
for example, responsible for throwing TypeError when you pass an object
that can not be either data descriptor or accessor one:

Object.defineProperty({}, 'x', { value: 'y', set: function(){} });
// or
Object.defineProperty({}, 'x', { get: function(){}, writable: true });

But then it might make sense to just look into relevant section of ES5
test suite on codeplex (https://es5conform.codeplex.com/).

[...]

--
kangax
First  |  Prev  | 
Pages: 1 2
Prev: Setter/Getter in module pattern?
Next: FAQ Topic - What is JScript? (2010-07-16)