First part of work to move V8 binding integrity off of vtables.

First part of work to move V8 binding integrity off of vtables.

Introduce a ScriptWrappable::init() templated method and call it from the ScriptWrappable sub-classes.

BUG=226633
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=147980

[Tom Sepez, 2013-04-08 22:48:30]

Adam, please take a look.

[abarth-chromium, 2013-04-08 22:50:55]

Did we discuss calling this function in the Foo::create method instead of the
constructor?

If A inherits from B, and B calls ScriptWrappable::init, does that cause a
problem for A if it wants to call ScriptWrappable::init too?

[Tom Sepez, 2013-04-08 22:56:50]

Yeah, we did, but there aren't always create methods, so this works even when
the object comes into existence in a strange manner.

Both A and B will call SW::init(), with the init from the most derived class
happening last, overwriting the subclass's call to SW::init.

[abarth-chromium, 2013-04-08 22:57:49]

Great.  LGTM

[Tom Sepez, 2013-04-08 22:58:00]

Both A and B will call SW::init(), with the init from the most derived class
happening last, overwriting the *superclass's* call to SW::init.

[abarth-chromium, 2013-04-08 22:59:08]

I presume you've run the layout tests in debug mode.  :)

[Tom Sepez, 2013-04-08 23:06:11]

On 2013/04/08 22:59:08, abarth wrote:
> I presume you've run the layout tests in debug mode.  :)
Yes.  A little noise, and 2 crashes on  my local linux64 system, neither of
which implicates scriptwrappables.  
But I just realized I sent it to the _rel trybots which is useless.

[Tom Sepez, 2013-04-09 22:22:00]

@abarth - this version neuters the assert in scriptwrappable that init was
called.  I will land that in a separate patch.  I'm getting some noise from the
trybots with that turned on, and with all these files, I'm pretty much running
as fast as I can to stay in one place wrt all the other rapid changes.

[Tom Sepez, 2013-04-09 22:26:59]

Committed patchset #4 manually as r147980 (presubmit successful).

[abarth-chromium, 2013-04-09 23:49:39]

Ok.

[esprehn, 2013-04-15 20:41:24]

What was the reason this wasn't just called in the initializer : SuperClass(..),
ScriptWrappable(this) ? Having to remember to call ::init() instead of making
the only constructor to ScriptWrappable be the one that takes the |this| would
catch bugs at compile time instead of the assert.

[Tom Sepez, 2013-04-15 20:47:47]

given:
ScriptWrappable <- class A <- class B, and we want to wrap both A and B with JS
different bindings, you need construct ScriptWrappable(B*) when constructing B,
which isn't possible when A has already constructed it as ScriptWrappable(A*).