IDL compiler: [Constructor] overloading

IDL compiler: [Constructor] overloading

This completes [Constructor].

Relatively long, b/c need to support list of constructors, but most of
logic already handled in method overloading.

A few notes:
* Also simplify Perl logic
* Some stub values (e.g., argument['is_nullable'] = False)
  b/c needed by overload resolution algorithm.
* moved constructor macro into methods.cpp b/c otherwise cyclic import
  between interface_base.cpp and interface.cpp.
* interface.cpp diff looks weird b/c it replaces the actual constructor
  method with the overload callback.

BUG=239771
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=163939

[Nils Barth (inactive), 2013-12-13 07:04:16]

Last [Constructor] CL!

https://codereview.chromium.org/112303003/diff/10001/Source/bindings/scripts/...
File Source/bindings/scripts/unstable/v8_interface.py (right):

https://codereview.chromium.org/112303003/diff/10001/Source/bindings/scripts/...
Source/bindings/scripts/unstable/v8_interface.py:350: 'is_nullable': False,
Stubs for overload resolution.

https://codereview.chromium.org/112303003/diff/10001/Source/bindings/template...
File Source/bindings/templates/interface.cpp (right):

https://codereview.chromium.org/112303003/diff/10001/Source/bindings/template...
Source/bindings/templates/interface.cpp:132: {% block
overloaded_constructor_callback %}
Diff looks weird; just look at new code.

This is b/c |constructor| block moved to macro in methods.cpp,
since needs to take a parameter (the constructor) and need to avoid
circular import,
and then new overload callback block added.

[haraken, 2013-12-13 10:23:31]

LGTM

> Last [Constructor] CL!

Each CL was small enough and very easy to review!

https://codereview.chromium.org/112303003/diff/10001/Source/bindings/scripts/...
File Source/bindings/scripts/unstable/v8_interface.py (right):

https://codereview.chromium.org/112303003/diff/10001/Source/bindings/scripts/...
Source/bindings/scripts/unstable/v8_interface.py:321: 'is_variadic': False,

Do we need to specify these, even though their values are hard-coded as True or
False? (Probably they're used in template files.)

https://codereview.chromium.org/112303003/diff/10001/Source/bindings/scripts/...
Source/bindings/scripts/unstable/v8_interface.py:360: if len(constructors) < 2:

Nit: I'd prefer '<= 1'

https://codereview.chromium.org/112303003/diff/10001/Source/bindings/template...
File Source/bindings/templates/methods.cpp (right):

https://codereview.chromium.org/112303003/diff/10001/Source/bindings/template...
Source/bindings/templates/methods.cpp:293:
throwTypeError(ExceptionMessages::failedToExecute("Constructor",
"{{interface_name}}",
ExceptionMessages::notEnoughArguments({{interface_length}}, info.Length())),
info.GetIsolate());

You can use the throw_type_error macro here, or you can remove the
throw_type_error macro.

https://codereview.chromium.org/112303003/diff/10001/Source/bindings/template...
Source/bindings/templates/methods.cpp:306: {% if
is_constructor_call_with_document %}

I hope you replace [CallWith=Document] with [CallWith=ExecutionContext] in the
near future.

https://codereview.chromium.org/112303003/diff/10001/Source/bindings/template...
Source/bindings/templates/methods.cpp:316:
V8DOMWrapper::associateObjectWithWrapper<{{v8_class}}>(impl.release(),
&{{v8_class}}::wrapperTypeInfo, wrapper, info.GetIsolate(),
WrapperConfiguration::Dependent);

Question: I wonder why we always specify 'Dependent'. We should specify
'Dependent' only for dependent objects (these objects have [DependentLifeTime]
or inherit from ActiveDOMObject) and specify 'Independent' for other objects.
Probably this is a bug of the IDL compiler.

[Nils Barth (inactive), 2013-12-16 03:15:56]

Thanks for quick reviews (hence rapid cycle);
revised and landing!

https://codereview.chromium.org/112303003/diff/10001/Source/bindings/scripts/...
File Source/bindings/scripts/unstable/v8_interface.py (right):

https://codereview.chromium.org/112303003/diff/10001/Source/bindings/scripts/...
Source/bindings/scripts/unstable/v8_interface.py:321: 'is_variadic': False,
On 2013/12/13 10:23:31, haraken wrote:
> Do we need to specify these, even though their values are hard-coded as True
or
> False? (Probably they're used in template files.)

Added a comment; it's required for overload resolution.
(Likewise for is_nullable and is_strict_type_checking.)

https://codereview.chromium.org/112303003/diff/10001/Source/bindings/scripts/...
Source/bindings/scripts/unstable/v8_interface.py:360: if len(constructors) < 2:
On 2013/12/13 10:23:31, haraken wrote:
> Nit: I'd prefer '<= 1'

n/p, done.

https://codereview.chromium.org/112303003/diff/10001/Source/bindings/template...
File Source/bindings/templates/methods.cpp (right):

https://codereview.chromium.org/112303003/diff/10001/Source/bindings/template...
Source/bindings/templates/methods.cpp:293:
throwTypeError(ExceptionMessages::failedToExecute("Constructor",
"{{interface_name}}",
ExceptionMessages::notEnoughArguments({{interface_length}}, info.Length())),
info.GetIsolate());
On 2013/12/13 10:23:31, haraken wrote:
> You can use the throw_type_error macro here,

Fixed; used macro consistently.

https://codereview.chromium.org/112303003/diff/10001/Source/bindings/template...
Source/bindings/templates/methods.cpp:306: {% if
is_constructor_call_with_document %}
On 2013/12/13 10:23:31, haraken wrote:
> I hope you replace [CallWith=Document] with [CallWith=ExecutionContext] in the
> near future.

Got it; will do after finish other constructors ([CustomConstructor] and
[NamedConstructor])!

https://codereview.chromium.org/112303003/diff/10001/Source/bindings/template...
Source/bindings/templates/methods.cpp:316:
V8DOMWrapper::associateObjectWithWrapper<{{v8_class}}>(impl.release(),
&{{v8_class}}::wrapperTypeInfo, wrapper, info.GetIsolate(),
WrapperConfiguration::Dependent);
On 2013/12/13 10:23:31, haraken wrote:
> Question: I wonder why we always specify 'Dependent'. We should specify
> 'Dependent' only for dependent objects (these objects have [DependentLifeTime]
> or inherit from ActiveDOMObject) and specify 'Independent' for other objects.
> Probably this is a bug of the IDL compiler.

Added a FIXME to that effect!

[commit-bot: I haz the power, 2013-12-16 03:25:10]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/nbarth@chromium.org/112303003/50001

[commit-bot: I haz the power, 2013-12-16 04:26:51]

Change committed as 163939