Have Node inherit EventTarget interface

Have Node inherit EventTarget interface

Have Node inherit EventTarget interface to match the specification:
http://dom.spec.whatwg.org/#interface-node

Node and its subclasses need slightly different implementation for
addEventListener() / removeEventListener(). Namely, we should not
call createHiddenDependency() / removeHiddenDependency() for Node
and its subclasses. To address this issue, a EventTarget::toNode()
virtual method is used so that we can make the check at runtime.

BUG=257583
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=153955

[haraken, 2013-07-10 12:24:35]

Don't we need to make some change to toEventTarget(), which is auto-generated
from the code generator when a given interface has [EvenTarget] ? (Rather, I
wonder why your previous CL that removed [EventTarget] from various interfaces
didn't cause problems around toEventTarget().)

[do-not-use, 2013-07-10 12:29:51]

On 2013/07/10 12:24:35, haraken wrote:
> Don't we need to make some change to toEventTarget(), which is auto-generated
> from the code generator when a given interface has [EvenTarget] ? (Rather, I
> wonder why your previous CL that removed [EventTarget] from various interfaces
> didn't cause problems around toEventTarget().)

I have been very careful to keep the [EventTarget] extended attribute on the
EventTarget interface until I update the bindings generator.
What the bindings generator does is 'InheritsExtendedAttribute($interface,
"EventTarget")'. 

The result of this check is unaffected by my patches because of the reason
above.

Once all the interfaces are updated to inherit EventTarget, we can then update
the generator to use InheritsInterface($interface, "EventTarget") instead of
InheritsExtendedAttribute($interface, "EventTarget").

[haraken, 2013-07-10 12:42:26]

On 2013/07/10 12:29:51, Christophe Dumez wrote:
> On 2013/07/10 12:24:35, haraken wrote:
> > Don't we need to make some change to toEventTarget(), which is
auto-generated
> > from the code generator when a given interface has [EvenTarget] ? (Rather, I
> > wonder why your previous CL that removed [EventTarget] from various
interfaces
> > didn't cause problems around toEventTarget().)
> 
> I have been very careful to keep the [EventTarget] extended attribute on the
> EventTarget interface until I update the bindings generator.
> What the bindings generator does is 'InheritsExtendedAttribute($interface,
> "EventTarget")'. 
> 
> The result of this check is unaffected by my patches because of the reason
> above.
> 
> Once all the interfaces are updated to inherit EventTarget, we can then update
> the generator to use InheritsInterface($interface, "EventTarget") instead of
> InheritsExtendedAttribute($interface, "EventTarget").

ah, makes sense.

LGTM, the change is fantastic. But please wait for an approval from arv or
abarth. I might be missing something.

[arv (Not doing code reviews), 2013-07-10 13:51:07]

LGTM

This makes me so happy.

[do-not-use, 2013-07-10 20:54:17]

Adam, does this one look OK to you as well?

[abarth-chromium, 2013-07-10 20:55:51]

LGTM.  There's obviously a compat risk, but we should try it.

[commit-bot: I haz the power, 2013-07-10 21:01:46]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/ch.dumez@sisa.samsung.com/18539006/16001

[do-not-use, 2013-07-10 21:18:05]

On 2013/07/10 20:55:51, abarth wrote:
> LGTM.  There's obviously a compat risk, but we should try it.

At least Firefox stable already has Node inheriting from EventTarget.

[arv (Not doing code reviews), 2013-07-10 21:22:51]

On 2013/07/10 21:18:05, Christophe Dumez wrote:
> On 2013/07/10 20:55:51, abarth wrote:
> > LGTM.  There's obviously a compat risk, but we should try it.
> 
> At least Firefox stable already has Node inheriting from EventTarget.

And IE since v9. This is low risk.

[abarth-chromium, 2013-07-10 23:56:46]

Great.

[commit-bot: I haz the power, 2013-07-11 01:50:52]

Change committed as 153955