[CSS Shapes] serialization of the computed value should omit the default radii

[CSS Shapes] serialization of the computed value should omit the default radii

We should omit closest-side value, wherever is possible. Specification:
​http://www.w3.org/TR/2014/CR-css-shapes-1-20140320/#basic-shape-computed-values
This change aligns the behavior to the specification.
Added new test cases for the shape-outside parsing tests.

This is a merge of http://trac.webkit.org/changeset/169406 by Zoltan Horvath <zoltan@webkit.org>;.

BUG=371934
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=175085

[rwlbuis, 2014-05-28 20:32:48]

PTAL :)

[Zoltan, 2014-05-28 21:30:13]

lgtm, thanks!

[rwlbuis, 2014-05-28 21:33:37]

PTAL, just a serialization fix.

[leviw_travelin_and_unemployed, 2014-05-28 22:07:41]

Okay. LGTM.

[eseidel, 2014-05-28 22:08:09]

https://codereview.chromium.org/300843011/diff/1/Source/core/css/CSSBasicShap...
File Source/core/css/CSSBasicShapes.cpp (right):

https://codereview.chromium.org/300843011/diff/1/Source/core/css/CSSBasicShap...
Source/core/css/CSSBasicShapes.cpp:118: if (m_radius && m_radius->getValueID()
!= CSSValueClosestSide)
This smells odd.  Why is CSSValueClosestSide special?  is that the default
value?  Can we access that in a lesss magic-constant way?

https://codereview.chromium.org/300843011/diff/1/Source/core/css/CSSBasicShap...
Source/core/css/CSSBasicShapes.cpp:190: bool shouldSerializeRadiusXValue =
m_radiusX->getValueID() != CSSValueClosestSide;
Here again.  Magic constants = confused reader.

[Zoltan, 2014-05-28 22:33:06]

On 2014/05/28 22:08:09, eseidel wrote:
>
https://codereview.chromium.org/300843011/diff/1/Source/core/css/CSSBasicShap...
> File Source/core/css/CSSBasicShapes.cpp (right):
> 
>
https://codereview.chromium.org/300843011/diff/1/Source/core/css/CSSBasicShap...
> Source/core/css/CSSBasicShapes.cpp:118: if (m_radius && m_radius->getValueID()
> != CSSValueClosestSide)
> This smells odd.  Why is CSSValueClosestSide special?  is that the default
> value?  Can we access that in a lesss magic-constant way?
> 
>
https://codereview.chromium.org/300843011/diff/1/Source/core/css/CSSBasicShap...
> Source/core/css/CSSBasicShapes.cpp:190: bool shouldSerializeRadiusXValue =
> m_radiusX->getValueID() != CSSValueClosestSide;
> Here again.  Magic constants = confused reader.

The spec says "omitting components when possible without changing the meaning". 
Based on the spec and its examples
(http://www.w3.org/TR/2014/CR-css-shapes-1-20140320/#basic-shape-serialization),
the closest-side should never be in the computed value and farthest-side should
always be there (when specified). That's why the 'magic' constant and that's why
it special. CSSValueClosestSide is a CSSValue which is accesible in
CSSBasicShapes.cpp, so I think it's convenient to use. 
What do you think?

[eseidel, 2014-05-29 15:37:57]

This still seems confusing to me, but you have a review from levi.

How would I know when I could change that code?  I don't know what those
constants mean or why radiusX should be closest-side?  If we had a concept of
CSS initial values (we do, but I think only for RenderStyle)
RenderStyle::radiusXInitial() would be much easier to read.

[rwlbuis, 2014-05-29 15:50:56]

On 2014/05/29 15:37:57, eseidel wrote:
> This still seems confusing to me, but you have a review from levi.

Thanks! We can always improve API, suggestions welcome :) In the mean time let's
get this in.

[rwlbuis, 2014-05-29 15:51:00]

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

[commit-bot: I haz the power, 2014-05-29 15:52:35]

CQ is trying da patch. Follow status at
 https://chromium-status.appspot.com/cq/rob.buis@samsung.com/300843011/1

[commit-bot: I haz the power, 2014-05-30 00:49:16]

Change committed as 175085