Bug 47922 – Merge ColorSpace and ImageColorSpace enums

• Component: SVG
• Status: RESOLVED FIXED

[Dirk Schulze]

I think we should merge the ColorSpace and ImageColorSpace enumerations. This mean adding linearRGB to ColorSpace and will be necessary for implementing color-interpolation as well as color-interpolation-filters on SVG.

[Dirk Schulze]

Created attachment 71198 [details]
Patch

[Nikolas Zimmermann]

Comment on attachment 71198 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=71198&action=review

Looks good overall, except for some minor things. But I'd really like to see whether we can avoid code duplication, by not using RetainPtr for the colorspace in ImageBuffer.

> WebCore/css/CSSPrimitiveValueMappings.h:2198
> +        default:

default is not added here on purpose, so the compiler warns if anything is missing. If you need to add sth. add the real case value.

> WebCore/platform/graphics/ColorSpace.h:34
> +    sRGBColorSpace,
> +    linearRGBColorSpace

I don't like these names, style guide says "Enum members should user InterCaps with an initial capital letter.".
So rename to SRGB and LinearRGB..

> WebCore/platform/graphics/cg/ImageBufferCG.cpp:83
> +    case DeviceColorSpace:
> +        m_data.m_colorSpace.adoptCF(CGColorSpaceCreateDeviceRGB());
> +        break;

Can you investigate, if it's possible to not use a RetainPtr for m_colorSpace. Then we could use the deviceRGBRef() static methods, instead of duplication code.

[Dirk Schulze]

Created attachment 71272 [details]
Patch

[Early Warning System Bot]

Attachment 71272 [details] did not build on qt:
Build output: http://queues.webkit.org/results/4516016

[Nikolas Zimmermann]

Comment on attachment 71272 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=71272&action=review

Other than that, looks good to me!

> WebCore/css/CSSPrimitiveValueMappings.h:2199
> +        ASSERT_NOT_REACHED();

You should comment here, that linearRgb is not available, through css "-webkit-color-correction". Otherwhise this is a bit confusing.

[Dirk Schulze]

Created attachment 71274 [details]
Patch

[Nikolas Zimmermann]

Comment on attachment 71274 [details]
Patch

Looks great, r=me.
Please do a leaks run first, as I'm not 100% certain about the removal of RetainPtr for m_colorSpace. Looks fine according to the CoreGraphics docs though, but better be safe than sorry.

[Dirk Schulze]

I checked. No leaks found. *Phew* ;-)

[Dirk Schulze]

Committed r70143: <http://trac.webkit.org/changeset/70143>

[WebKit Review Bot]

http://trac.webkit.org/changeset/70143 might have broken Qt Linux Release
The following tests are not passing:
http/tests/security/mixedContent/insecure-css-in-main-frame.html