Refactor CSSGradientValue

Refactor CSSGradientValue

After r186438 currentColor is evaluated when the actual
background image is created. This means m_resolvedColor
and gradientWithStylesResolved can be removed when in
other code spots we resolve the color using the passed
RenderObject.

Also with gradientWithStylesResolved gone some call sites
can be simplified.

Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=186701

[rwlbuis, 2014-12-04 23:48:33]

rob.buis@samsung.com changed reviewers:
+ timloh@chromium.org

[rwlbuis, 2014-12-04 23:48:33]

PTAL, WDYT?

[Timothy Loh, 2014-12-05 00:07:07]

This is probably fine. I still find it weird that gradients don't resolve their
styles (other properties would store a Length for position and a StyleColor for
color), but it probably isn't worth trying to make this change.

https://codereview.chromium.org/778723002/diff/100001/Source/core/css/CSSGrad...
File Source/core/css/CSSGradientValue.h (left):

https://codereview.chromium.org/778723002/diff/100001/Source/core/css/CSSGrad...
Source/core/css/CSSGradientValue.h:62: bool m_colorIsDerivedFromElement;
You'd probably want to update usage of this too?

[rwlbuis, 2014-12-05 17:04:01]

On 2014/12/05 00:07:07, Timothy Loh wrote:
> This is probably fine. I still find it weird that gradients don't resolve
their
> styles (other properties would store a Length for position and a StyleColor
for
> color), but it probably isn't worth trying to make this change.
> 
>
https://codereview.chromium.org/778723002/diff/100001/Source/core/css/CSSGrad...
> File Source/core/css/CSSGradientValue.h (left):
> 
>
https://codereview.chromium.org/778723002/diff/100001/Source/core/css/CSSGrad...
> Source/core/css/CSSGradientValue.h:62: bool m_colorIsDerivedFromElement;
> You'd probably want to update usage of this too?

Done!

[Timothy Loh, 2014-12-07 23:20:43]

lgtm

[andersr, 2014-12-08 12:13:42]

On 2014/12/05 00:07:07, Timothy Loh wrote:
> This is probably fine. I still find it weird that gradients don't resolve
their
> styles (other properties would store a Length for position and a StyleColor
for
> color), but it probably isn't worth trying to make this change.

Because Lengths aren't properly resolved during style resolution for generated
images, there should theoretically be a bug related to viewport units and style
recalc, where the Document might not know that one of its RenderStyles is using
viewport units. So it's worse than just weird.

[rwlbuis, 2014-12-08 13:43:09]

The CQ bit was checked by rob.buis@samsung.com

[commit-bot: I haz the power, 2014-12-08 13:43:18]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/778723002/120001

[commit-bot: I haz the power, 2014-12-08 14:51:14]

Committed patchset #7 (id:120001) as
https://src.chromium.org/viewvc/blink?view=rev&revision=186701