IDL: Properly support [Exposed] on static methods

IDL: Properly support [Exposed] on static methods

They were conditionally installed as indicated by the [Exposed] attribute,
but they were installed on the prototype object rather than on the
interface object.

A static method with [Exposed] fulfills the requirements for being treated
both as a conditionally enabled method (because of [Exposed]) and a method
with custom registration (because of being static), and due to the order
in which those requirements were checked, it ended up being treated as a
conditionally enabled method, only.

Fix by giving custom registration priority and by adding the same
condition filters around that registration code, so that such methods can
still also be all sorts of conditionally enabled.

BUG=434741
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=187161

[Jens Widell, 2014-12-15 12:14:00]

jl@opera.com changed reviewers:
+ haraken@chromium.org

[Jens Widell, 2014-12-15 12:14:00]

PTAL

[Jens Widell, 2014-12-15 12:22:35]

https://codereview.chromium.org/793243004/diff/1/Source/bindings/scripts/v8_i...
File Source/bindings/scripts/v8_interface.py (right):

https://codereview.chromium.org/793243004/diff/1/Source/bindings/scripts/v8_i...
Source/bindings/scripts/v8_interface.py:390:
custom_registration_methods.append(method)
So, I split 'has_custom_registration' away from 'runtime_enabled_function' here,
so that the former takes priority over 'per_context_enabled_function' and
'conditionally_exposed_function' but not the latter.

If I had just changed the order of the existing if statements, runtime enabled
*and* per context enabled or conditionally exposed methods would have moved from
conditionally enabled (installConditionallyEnabledMethods()) to custom
registration. The primary difference, I think, is that the former does
prototypeObject->Set() and the latter uses V8DOMConfiguration::installMethod()
and installs the method on the prototype template. I'm not entirely sure what
the difference between those two alternatives is in practice.

Anyway, doing it this way minimizes impact. Current real IDL have no methods
affected by this change, so no CG changes. Test IDL files have some
conditionally exposed static methods that are now treated differently
(correctly) and that's the only thing that changes. If I had changed the order
of the if statement, quite a lot of more things would have changed, both in the
tests and the real generated code.

[Jens Widell, 2014-12-15 12:32:24]

https://codereview.chromium.org/793243004/diff/1/Source/bindings/tests/result...
File Source/bindings/tests/results/core/V8TestInterface.cpp (right):

https://codereview.chromium.org/793243004/diff/1/Source/bindings/tests/result...
Source/bindings/tests/results/core/V8TestInterface.cpp:2086: if (context &&
context->isDocument() &&
ContextFeatures::partialContextName3Enabled(toDocument(context))) {
This added condition is wrong. Only one of the overloads is per-context enabled,
but now the whole method is. (Worth pointing out though that previously we
simply ignored the method's [PerContextEnabled] completely.)

[Jens Widell, 2014-12-15 12:37:55]

https://codereview.chromium.org/793243004/diff/1/Source/bindings/tests/result...
File Source/bindings/tests/results/core/V8TestInterface.cpp (right):

https://codereview.chromium.org/793243004/diff/1/Source/bindings/tests/result...
Source/bindings/tests/results/core/V8TestInterface.cpp:2086: if (context &&
context->isDocument() &&
ContextFeatures::partialContextName3Enabled(toDocument(context))) {
On 2014/12/15 12:32:24, Jens Widell wrote:
> This added condition is wrong. Only one of the overloads is per-context
enabled,
> but now the whole method is. (Worth pointing out though that previously we
> simply ignored the method's [PerContextEnabled] completely.)

Fixed in PS#2.

[haraken, 2014-12-15 15:13:54]

LGTM

[Jens Widell, 2014-12-15 15:47:34]

The CQ bit was checked by jl@opera.com

[commit-bot: I haz the power, 2014-12-15 15:47:48]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/793243004/20001

[commit-bot: I haz the power, 2014-12-15 18:30:35]

Committed patchset #2 (id:20001) as
https://src.chromium.org/viewvc/blink?view=rev&revision=187161

[samuong, 2014-12-15 21:40:36]

A revert of this CL (patchset #2 id:20001) has been created in
https://codereview.chromium.org/806773006/ by samuong@chromium.org.

The reason for reverting is: seems to be causing compile failures:
http://build.chromium.org/p/chromium.webkit/builders/WebKit%20Mac%20Builder%2....