Bug 192480 – CSS Painting API code cleanup

• Component: Layout and Rendering
• Status: RESOLVED FIXED

[Justin Michaud]

The goal of this is to fix all of Darin's comments on 191443, plus a few other housekeeping things.

[Justin Michaud]

Created attachment 356763 [details]
Patch

[WebKit Commit Bot]

Comment on attachment 356763 [details]
Patch

Clearing flags on attachment: 356763

Committed r238968: <https://trac.webkit.org/changeset/238968>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.

[Radar WebKit Bug Importer]

<rdar://problem/46564091>

[Simon Fraser (smfr)]

Comment on attachment 356763 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=356763&action=review

> Source/WebCore/rendering/style/RenderStyle.cpp:1042
> +    if (changedCustomPaintWatchedProperty(*this, *m_rareNonInheritedData, other, *other.m_rareNonInheritedData))
> +        return true;

I don't think this is the right way to do this. You shouldn't trigger repaint for custom paint from here. You should evaluate inputs to custom paints, run the callback if necessary, then issue a repaint if the output changes.

[Justin Michaud]

https://bugs.webkit.org/show_bug.cgi?id=192322

[Justin Michaud]

Comment on attachment 356763 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=356763&action=review

>> Source/WebCore/rendering/style/RenderStyle.cpp:1042
>> +        return true;
> 
> I don't think this is the right way to do this. You shouldn't trigger repaint for custom paint from here. You should evaluate inputs to custom paints, run the callback if necessary, then issue a repaint if the output changes.

I have a test case that seems to prove you right. If you mutate an a watched property immediately after importing a worklet, then it does not paint correctly.