IDL: fix code generation for attributes with [PutForwards]

IDL: fix code generation for attributes with [PutForwards]

When an attribute forwards puts to another attribute, most details in the
generated setter code depend on the target (forwarded-to) attribute's type
and attributes rather than on those of the source attribute (that is being
forwarded.)

If the target attribute (or its interface) has [TypeChecking=Interface]
and an interface type, the value set should be checked against that
interface type. Likewise, if the source attribute (or its interface) has
[TypeChecking=Interface] and an interface type, the value set should *not*
be checked against it, since that's not what is being set.

For instance, given

  [TypeChecking=Interface]
  interface Document {
    [PutForwards=href] attribute Location location;
  }

the setter should not check that the value to set is a Location object,
since it will actually be converted to a string and assigned to the
location object's href attribute, which is a DOMString attribute.

Also, whether the setter needs an ExceptionState also depends on the
target attribute's type and [RaisesException], not the source attribute's,
since it's the conversion to the target attribute's type and/or the target
attribute's actual setter that would be needing (or not) the exception
state.

Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=176985

[Jens Widell, 2014-06-26 10:48:05]

PTAL

I'm mostly moving setter-specific code from attribute_context() into
setter_context(), where it could be argued it makes sense to have
setter-specific code.

This was triggered by a real-life problem: I tried adding
[TypeChecking=Interface] to the Document interface, and that broke the
Document.location setter.

These changes don't affect code generation with current IDL files (other than
those in bindings/tests/idls/, of course.)

[haraken, 2014-06-26 12:02:24]

LGTM

https://codereview.chromium.org/353973002/diff/40001/Source/bindings/scripts/...
File Source/bindings/scripts/v8_attributes.py (right):

https://codereview.chromium.org/353973002/diff/40001/Source/bindings/scripts/...
Source/bindings/scripts/v8_attributes.py:313: context['cpp_setter'] =
setter_expression(interface, attribute, context)

Shall we put this into the above context.update() ?

[Jens Widell, 2014-06-26 12:11:47]

https://codereview.chromium.org/353973002/diff/40001/Source/bindings/scripts/...
File Source/bindings/scripts/v8_attributes.py (right):

https://codereview.chromium.org/353973002/diff/40001/Source/bindings/scripts/...
Source/bindings/scripts/v8_attributes.py:313: context['cpp_setter'] =
setter_expression(interface, attribute, context)
On 2014/06/26 12:02:24, haraken wrote:
> 
> Shall we put this into the above context.update() ?

setter_expression() uses context['is_setter_raises_exception'], which is also
set in the above context.update(), so that doesn't work. But we can of course
pass in 'is_setter_raises_exception' to it as an argument instead.

It's a bit unfortunate I suppose that setter_expression() both depends on
context's contents and updates it, while returning an expression only to have
that too stored in the context.

[haraken, 2014-06-26 12:22:37]

On 2014/06/26 12:11:47, Jens Lindström wrote:
>
https://codereview.chromium.org/353973002/diff/40001/Source/bindings/scripts/...
> File Source/bindings/scripts/v8_attributes.py (right):
> 
>
https://codereview.chromium.org/353973002/diff/40001/Source/bindings/scripts/...
> Source/bindings/scripts/v8_attributes.py:313: context['cpp_setter'] =
> setter_expression(interface, attribute, context)
> On 2014/06/26 12:02:24, haraken wrote:
> > 
> > Shall we put this into the above context.update() ?
> 
> setter_expression() uses context['is_setter_raises_exception'], which is also
> set in the above context.update(), so that doesn't work. But we can of course
> pass in 'is_setter_raises_exception' to it as an argument instead.
> 
> It's a bit unfortunate I suppose that setter_expression() both depends on
> context's contents and updates it, while returning an expression only to have
> that too stored in the context.

ok, it makes sense to keep it outside context.update().

[Jens Widell, 2014-06-26 12:26:19]

The CQ bit was checked by jl@opera.com

[commit-bot: I haz the power, 2014-06-26 12:27:19]

CQ is trying da patch. Follow status at
 https://chromium-status.appspot.com/cq/jl@opera.com/353973002/40001

[commit-bot: I haz the power, 2014-06-26 15:14:53]

Change committed as 176985