Bug 61449 – Update RGBA color serialization to match HTML5

• Component: WebCore Misc.
• Status: RESOLVED FIXED

[Andreas Kling]

As of this change:

http://html5.org/tools/web-apps-tracker?from=5874&to=5875

HTML5 now specifies that zero alpha values should be serialized to "0". We currently yield "0.0" or "0." followed by 5 decimals to match Gecko, but Firefox nightlies are now returning "0" or "0." followed by the least number of decimals necessary.

[Andreas Kling]

Created attachment 94807 [details]
Proposed patch

[Benjamin Poulain]

Comment on attachment 94807 [details]
Proposed patch

View in context: https://bugs.webkit.org/attachment.cgi?id=94807&action=review

> Source/WebCore/platform/graphics/Color.cpp:211
> +    if (!hasAlpha())
> +        result.append('1');
> +    else if (!alpha())

I don't know how often this is called (or if the compiler does a good job here) so maybe that is not relevant.
But this will call alpha() twice: for hasAlpha() and second check for hasAlpha() :)

Also I am not a fan of "hasAlpha()" since it is not clear if that mean alpha == 0 or alpha == 1 :)
"alpha() == 255" is clearer, at least for me.

[Andreas Kling]

Created attachment 94810 [details]
Proposed patch v2

[Andreas Kling]

Created attachment 94811 [details]
Proposed patch v2

[WebKit Review Bot]

Comment on attachment 94811 [details]
Proposed patch v2

Attachment 94811 [details] did not pass chromium-ews (chromium-xvfb):
Output: http://queues.webkit.org/results/8739035

New failing tests:
editing/style/inline-style-container.html

[WebKit Review Bot]

Created attachment 94918 [details]
Archive of layout-test-results from ec2-cr-linux-01

The attached test failures were seen while running run-webkit-tests on the chromium-ews.
Bot: ec2-cr-linux-01  Port: Chromium  Platform: Linux-2.6.35-28-virtual-x86_64-with-Ubuntu-10.10-maverick

[Andreas Kling]

(In reply to comment #5)
> (From update of attachment 94811 [details])
> Attachment 94811 [details] did not pass chromium-ews (chromium-xvfb):
> Output: http://queues.webkit.org/results/8739035
> 
> New failing tests:
> editing/style/inline-style-container.html

Need a minor expectation tweak in editing/style/inline-style-container.html, will fix when landing if r+ed.

[Alexis Menard (darktears)]

Comment on attachment 94811 [details]
Proposed patch v2

LGTM modulo the breakage :D

[Benjamin Poulain]

Comment on attachment 94811 [details]
Proposed patch v2

String::format("%.5f", alpha() / 255.0f)); -> DecimalNumber(alpha() / 255.0) will create different values regarding float/double and the rounding of float/dtoa.
But as far as I can see, that is for the better, so r+ :)

[Andreas Kling]

Committed r87512: <http://trac.webkit.org/changeset/87512>