Use IDL union types for CanvasRenderingContext2D.{fill,stroke}Style

Use IDL union types for CanvasRenderingContext2D.{fill,stroke}Style

This gets rid of the custom bindings for said attributes by using the
union type support in the bindings codegen.
Also rework the resolution of 'currentColor' by using
parseColorOrCurrentColor directly in set{Fill,Stroke}Style, rather than
first parse and then resolve.
This allows CanvasStyle to cleaned up and shrunk a bit.
Also remove code from CanvasStyle that became dead when the legacy
set{Fill,Stroke}Color methods was removed (CMYKA, float RGBA,
gray-level, override alpha).

BUG=430337
BUG=363180, 363181
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=185879

[fs, 2014-11-24 13:20:23]

fs@opera.com changed reviewers:
+ bashi@chromium.org, jl@opera.com, junov@chromium.org

[fs, 2014-11-24 13:20:25]

https://codereview.chromium.org/752083003/diff/1/Source/core/html/canvas/Canv...
File Source/core/html/canvas/CanvasRenderingContext2D.cpp (right):

https://codereview.chromium.org/752083003/diff/1/Source/core/html/canvas/Canv...
Source/core/html/canvas/CanvasRenderingContext2D.cpp:463: if (colorString ==
state().m_unparsedStrokeColor)
The removal of the legacy APIs seems to have dropped this optimization, but left
the state-field, so I recovered it. If it was intentional to remove it, then I
can drop this (and the state-field).

https://codereview.chromium.org/752083003/diff/1/Source/core/html/canvas/Canv...
File Source/core/html/canvas/CanvasRenderingContext2D.idl (left):

https://codereview.chromium.org/752083003/diff/1/Source/core/html/canvas/Canv...
Source/core/html/canvas/CanvasRenderingContext2D.idl:66: // FIXME: Use union
types when supported: http://crbug.com/372891
This bug seems to have some sort Restrict* label set (or at least I couldn't
access it).

https://codereview.chromium.org/752083003/diff/1/Source/core/html/canvas/Canv...
File Source/core/html/canvas/CanvasStyle.h (right):

https://codereview.chromium.org/752083003/diff/1/Source/core/html/canvas/Canv...
Source/core/html/canvas/CanvasStyle.h:61: enum Type { ColorRGBA, Gradient,
ImagePattern };
Renamed this to make presubmit happy.

[Jens Widell, 2014-11-24 13:38:24]

Union type usage LGTM, and custom bindings removal LGTM++.

https://codereview.chromium.org/752083003/diff/1/Source/core/html/canvas/Canv...
File Source/core/html/canvas/CanvasRenderingContext2D.idl (left):

https://codereview.chromium.org/752083003/diff/1/Source/core/html/canvas/Canv...
Source/core/html/canvas/CanvasRenderingContext2D.idl:66: // FIXME: Use union
types when supported: http://crbug.com/372891
On 2014/11/24 13:20:24, fs wrote:
> This bug seems to have some sort Restrict* label set (or at least I couldn't
> access it).

That's an unrelated crash bug (CC:ed oilpan-reviews@, so I could randomly see
it.)

The bug that ought to have been referenced here is: http://crbug.com/240176

[fs, 2014-11-24 13:49:07]

https://codereview.chromium.org/752083003/diff/1/Source/core/html/canvas/Canv...
File Source/core/html/canvas/CanvasRenderingContext2D.idl (left):

https://codereview.chromium.org/752083003/diff/1/Source/core/html/canvas/Canv...
Source/core/html/canvas/CanvasRenderingContext2D.idl:66: // FIXME: Use union
types when supported: http://crbug.com/372891
On 2014/11/24 13:38:24, Jens Widell wrote:
> On 2014/11/24 13:20:24, fs wrote:
> > This bug seems to have some sort Restrict* label set (or at least I couldn't
> > access it).
> 
> That's an unrelated crash bug (CC:ed oilpan-reviews@, so I could randomly see
> it.)
> 
> The bug that ought to have been referenced here is: http://crbug.com/240176

Thanks, that's the info I was probing for. As discussed elsewhere, I'll replace
the ref to the above (240176) with 430337.

[haraken, 2014-11-24 14:47:23]

haraken@chromium.org changed reviewers:
+ haraken@chromium.org

[haraken, 2014-11-24 14:47:23]

LGTM

[Justin Novosad, 2014-11-24 15:41:02]

lgtm

https://codereview.chromium.org/752083003/diff/1/Source/core/html/canvas/Canv...
File Source/core/html/canvas/CanvasRenderingContext2D.cpp (right):

https://codereview.chromium.org/752083003/diff/1/Source/core/html/canvas/Canv...
Source/core/html/canvas/CanvasRenderingContext2D.cpp:463: if (colorString ==
state().m_unparsedStrokeColor)
On 2014/11/24 13:20:24, fs wrote:
> The removal of the legacy APIs seems to have dropped this optimization, but
left
> the state-field, so I recovered it. If it was intentional to remove it, then I
> can drop this (and the state-field).

Thanks for catching that. the removal was unintentional.

[fs, 2014-11-24 15:59:37]

The CQ bit was checked by fs@opera.com

[commit-bot: I haz the power, 2014-11-24 16:00:43]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/752083003/1

[commit-bot: I haz the power, 2014-11-24 16:07:08]

Committed patchset #1 (id:1) as
https://src.chromium.org/viewvc/blink?view=rev&revision=185879