bindings: Makes the default receiver of callbacks "undefined".

bindings: Makes the default receiver of callbacks "undefined".

http://heycam.github.io/webidl/#es-invoking-callback-functions
By default, "the callback this value" should be "undefined".

In strict mode,
- before: this value was window (global object).
- after: this value is undefined

In non-strict mode,
undefined (or null) this value is converted to the global object.
This conversion is done in V8.  So user script observes no change.

See http://www.ecma-international.org/ecma-262/5.1/#sec-10.4.3

Note: This CL changes the callback this value from the window object of the receiver to the global object of the running script.
Example) Suppose running the following script in a child window.
    function f() { console.log(this); }
    top.requestAnimationFrame(f);
before: this === top
after: this === window (child window)

BUG=332663
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=201482

[Yuki, 2015-08-31 07:19:18]

yukishiino@chromium.org changed reviewers:
+ haraken@chromium.org

[Yuki, 2015-08-31 07:19:19]

Could you review this CL?

[haraken, 2015-08-31 07:33:09]

On 2015/08/31 07:19:19, Yuki wrote:
> Could you review this CL?

Would you describe the latest behavior of Firefox, Safari and IE? From the bug
description, it seems that Firefox is "sort of PASS".

[Jens Widell, 2015-08-31 07:35:46]

jl@opera.com changed reviewers:
+ jl@opera.com

[Jens Widell, 2015-08-31 07:35:46]

LGTM

One thing I'd check is whether which global object becomes this changes in a
multiple-document scenario. E.g., if you have a page in a frame, and you do

  <script>
    function f() { alert(this.location.href); }
    top.requestAnimationFrame(f);
  </script>

Will you get the same result with this patch as before? (And if not, is the new
behavior better? I'd guess it was, if so.)

[Yuki, 2015-08-31 10:17:04]

Firefox supports "use strict" and |this| becomes undefined.
IE doesn't support it well and |this| becomes window.
I don't have Safari at this moment.

On 2015/08/31 07:35:46, Jens Widell wrote:
> One thing I'd check is whether which global object becomes this changes in a
> multiple-document scenario. E.g., if you have a page in a frame, and you do
> 
>   <script>
>     function f() { alert(this.location.href); }
>     top.requestAnimationFrame(f);
>   </script>
> 
> Will you get the same result with this patch as before? (And if not, is the
new
> behavior better? I'd guess it was, if so.)

That's a very good point.

Old Blink: this => the top window
This patch: this => the global object of the script is running, i.e. child
window.
Firefox: this => child window, the same as the new behavior of Blink.

This CL changes the behavior, but I think the new behavior is correct because
"the global object" is child window where/when the script is running.  So it
should be the child window.

What do you think?

[haraken, 2015-08-31 10:32:04]

On 2015/08/31 10:17:04, Yuki wrote:
> Firefox supports "use strict" and |this| becomes undefined.
> IE doesn't support it well and |this| becomes window.
> I don't have Safari at this moment.
> 
> On 2015/08/31 07:35:46, Jens Widell wrote:
> > One thing I'd check is whether which global object becomes this changes in a
> > multiple-document scenario. E.g., if you have a page in a frame, and you do
> > 
> >   <script>
> >     function f() { alert(this.location.href); }
> >     top.requestAnimationFrame(f);
> >   </script>
> > 
> > Will you get the same result with this patch as before? (And if not, is the
> new
> > behavior better? I'd guess it was, if so.)
> 
> That's a very good point.
> 
> Old Blink: this => the top window
> This patch: this => the global object of the script is running, i.e. child
> window.
> Firefox: this => child window, the same as the new behavior of Blink.
> 
> This CL changes the behavior, but I think the new behavior is correct because
> "the global object" is child window where/when the script is running.  So it
> should be the child window.
> 
> What do you think?

Sounds reasonable to me.

LGTM.

[Jens Widell, 2015-08-31 10:45:59]

On 2015/08/31 10:17:04, Yuki wrote:
> Old Blink: this => the top window
> This patch: this => the global object of the script is running, i.e. child
> window.
> Firefox: this => child window, the same as the new behavior of Blink.
> 
> This CL changes the behavior, but I think the new behavior is correct because
> "the global object" is child window where/when the script is running.  So it
> should be the child window.
> 
> What do you think?

The new behavior is what I'd expect, so I think this is an all-around good
change.

[Yuki, 2015-08-31 11:59:26]

The CQ bit was checked by yukishiino@chromium.org

[commit-bot: I haz the power, 2015-08-31 11:59:35]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1301423004/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1301423004/1

[commit-bot: I haz the power, 2015-08-31 12:50:29]

Committed patchset #1 (id:1) as
https://src.chromium.org/viewvc/blink?view=rev&revision=201482