Bug 161855 – Add ref-counted ColorImpl to Color

• Status: RESOLVED
• Resolution: WONTFIX
• Component: Layout and Rendering

[Dean Jackson]

As part of https://bugs.webkit.org/show_bug.cgi?id=160520 : Add a RefPtr to a ColorImpl from Color.

This, by itself, is useless, so it won't land in isolation.

[Radar WebKit Bug Importer]

<rdar://problem/28253890>

[Dean Jackson]

Created attachment 288547 [details]
Patch

[Dean Jackson]

Created attachment 288549 [details]
Patch

[WebKit Commit Bot]

Attachment 288549 [details] did not pass style-queue:


ERROR: Source/WebCore/ChangeLog:16:  Need whitespace between colon and description  [changelog/filechangedescriptionwhitespace] [5]
Total errors found: 1 in 8 files


If any of these errors are false positives, please file a bug against check-webkit-style.

[Dean Jackson]

Created attachment 288627 [details]
Patch

[Simon Fraser (smfr)]

Comment on attachment 288627 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=288627&action=review

> Source/WebCore/platform/graphics/Color.h:196
>      RGBA32 m_color;
>      bool m_valid;
> +    RefPtr<ColorImpl> m_colorImpl;

Color is 16 bytes with this layout, but would be 13 bytes if you put the RefPtr first.

[Dean Jackson]

Comment on attachment 288627 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=288627&action=review

>> Source/WebCore/platform/graphics/Color.h:196
>> +    RefPtr<ColorImpl> m_colorImpl;
> 
> Color is 16 bytes with this layout, but would be 13 bytes if you put the RefPtr first.

This doesn't matter since m_color and m_valid will go away, and Color will end up as 8 bytes.

I'm splitting the patches into small chunks to make it reasonable. This part will only land when all the component pieces are ready, so that we won't ever be in a situation with a 16 byte Color.

[Dean Jackson]

Now done a better way in https://bugs.webkit.org/show_bug.cgi?id=162878