Remove Constructor attributes on DOMWindow because of [EnabledAtRuntime]

Remove Constructor attributes on DOMWindow because of [EnabledAtRuntime]

Add [EnabledAtRuntime] to several interfaces and remove [NoInterfaceObject] so
that their global constructors are now automatically generated and enabled at
runtime. We no longer need explicit attributes for those interfaces on
DOMWindow interface.

DeviceMotionEvent tests are now skipped because the interface is now disabled
at runtime, not just its constructor.

Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=152127

[do-not-use, 2013-06-10 09:58:35]

https://codereview.chromium.org/16691003/diff/2001/LayoutTests/TestExpectations
File LayoutTests/TestExpectations (right):

https://codereview.chromium.org/16691003/diff/2001/LayoutTests/TestExpectatio...
LayoutTests/TestExpectations:118: # Unskip when DeviceMotion is implemented and
enabled at runtime.
Alternatively, we could rebaseline those. This seems to be the approach chosen
for fast/dom/DeviceMotion/window-property.html which fails because the window
constructor is disabled at runtime.

[haraken, 2013-06-10 10:14:55]

https://codereview.chromium.org/16691003/diff/2001/LayoutTests/TestExpectations
File LayoutTests/TestExpectations (right):

https://codereview.chromium.org/16691003/diff/2001/LayoutTests/TestExpectatio...
LayoutTests/TestExpectations:118: # Unskip when DeviceMotion is implemented and
enabled at runtime.

Why does this CL change the behavior? I expected this CL doesn't change
generated code.

[do-not-use, 2013-06-10 10:26:57]

On 2013/06/10 10:14:55, haraken wrote:
>
https://codereview.chromium.org/16691003/diff/2001/LayoutTests/TestExpectations
> File LayoutTests/TestExpectations (right):
> 
>
https://codereview.chromium.org/16691003/diff/2001/LayoutTests/TestExpectatio...
> LayoutTests/TestExpectations:118: # Unskip when DeviceMotion is implemented
and
> enabled at runtime.
> 
> Why does this CL change the behavior? I expected this CL doesn't change
> generated code.

Having [EnabledAtRuntime] on an interface has an impact on the generated code
(besides the generated constructor attributes). From CodeGeneratorV8.pl:
    if ($interface->extendedAttributes->{"EnabledAtRuntime"}) {
        my $enable_function = GetRuntimeEnableFunctionName($interface);
        $code .= <<END;
    if (!${enable_function}())
        defaultSignature = V8DOMConfiguration::configureTemplate(desc, \"\",
$parentClassTemplate, ${v8ClassName}::internalFieldCount, 0, 0, 0, 0, isolate,
currentWorldType);
    else
END
    }

The DeviceMotionEvent was already disabled at runtime. However, the tests are
creating a DeviceMotionEvent using document.createEvent('DeviceMotionEvent');
which was working until now.

[do-not-use, 2013-06-10 10:27:43]

On 2013/06/10 10:26:57, Christophe Dumez wrote:
> On 2013/06/10 10:14:55, haraken wrote:
> >
>
https://codereview.chromium.org/16691003/diff/2001/LayoutTests/TestExpectations
> > File LayoutTests/TestExpectations (right):
> > 
> >
>
https://codereview.chromium.org/16691003/diff/2001/LayoutTests/TestExpectatio...
> > LayoutTests/TestExpectations:118: # Unskip when DeviceMotion is implemented
> and
> > enabled at runtime.
> > 
> > Why does this CL change the behavior? I expected this CL doesn't change
> > generated code.
> 
> Having [EnabledAtRuntime] on an interface has an impact on the generated code
> (besides the generated constructor attributes). From CodeGeneratorV8.pl:
>     if ($interface->extendedAttributes->{"EnabledAtRuntime"}) {
>         my $enable_function = GetRuntimeEnableFunctionName($interface);
>         $code .= <<END;
>     if (!${enable_function}())
>         defaultSignature = V8DOMConfiguration::configureTemplate(desc, \"\",
> $parentClassTemplate, ${v8ClassName}::internalFieldCount, 0, 0, 0, 0, isolate,
> currentWorldType);
>     else
> END
>     }
> 
> The DeviceMotionEvent was already disabled at runtime. However, the tests are
> creating a DeviceMotionEvent using document.createEvent('DeviceMotionEvent');
> which was working until now.

I meant "The DeviceMotionEvent *constructor* was already disabled..."

[haraken, 2013-06-10 10:37:09]

Thanks for clarification. Makes sense. LGTM.

For the time being, just skipping the tests sounds reasonable to me. A right way
to fix the issue would be implementing a testing mechanism for a
[EnabledAtRuntime] world. We should have it.

[commit-bot: I haz the power, 2013-06-10 11:20:34]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/ch.dumez@sisa.samsung.com/16691003/2001

[commit-bot: I haz the power, 2013-06-10 12:35:59]

Change committed as 152127

[arv (Not doing code reviews), 2013-06-10 13:44:27]

LGTM