compositor-worker: plumb element id and mutable properties

third_party/WebKit/Source/core/css/resolver/StyleAdjuster.cpp

 /*
  * Copyright (C) 1999 Lars Knoll (knoll@kde.org)
  *           (C) 2004-2005 Allan Sandfeld Jensen (kde@carewolf.com)
  * Copyright (C) 2006, 2007 Nicholas Shanks (webkit@nickshanks.com)
  * Copyright (C) 2005, 2006, 2007, 2008, 2009, 2010, 2011, 2012, 2013 Apple Inc. All rights reserved.
  * Copyright (C) 2007 Alexey Proskuryakov <ap@webkit.org>
  * Copyright (C) 2007, 2008 Eric Seidel <eric@webkit.org>
  * Copyright (C) 2008, 2009 Torch Mobile Inc. All rights reserved. (http://www.torchmobile.com/)
  * Copyright (c) 2011, Code Aurora Forum. All rights reserved.
  * Copyright (C) Research In Motion Limited 2011. All rights reserved.
@@ skipping 31 matching lines @@
 #include "core/html/HTMLPlugInElement.h"
 #include "core/html/HTMLTableCellElement.h"
 #include "core/html/HTMLTextAreaElement.h"
 #include "core/layout/LayoutReplaced.h"
 #include "core/layout/LayoutTheme.h"
 #include "core/style/ComputedStyle.h"
 #include "core/style/ComputedStyleConstants.h"
 #include "core/svg/SVGSVGElement.h"
 #include "platform/Length.h"
 #include "platform/transforms/TransformOperations.h"
+#include "public/platform/WebCompositorMutableProperties.h"
 #include "wtf/Assertions.h"
 
 namespace blink {
 
 using namespace HTMLNames;
 
 static EDisplay equivalentBlockDisplay(EDisplay display, bool isFloating, bool strictParsing)
 {
     switch (display) {
     case BLOCK:
@@ skipping 112 matching lines @@
 
         // We don't adjust the first letter style earlier because we may change the display setting in
         // adjustStyeForTagName() above.
         adjustStyleForFirstLetter(style);
 
         adjustStyleForDisplay(style, parentStyle, element ? &element->document() : 0);
     } else {
         adjustStyleForFirstLetter(style);
     }
 
+    if (element && element->hasCompositorProxy()) {

[chrishtr, 2015/11/24 16:22:50]

Factor into another method, this one is getting too long.

[Ian Vollick, 2015/11/25 18:28:44]

Done.

+        TRACE_EVENT0(TRACE_DISABLED_BY_DEFAULT("compositor-worker"), "StyleAdjuster::adjustComputedStyle hasCompositorProxy");
+
+        style.setHasCompositorProxy(true);
+        uint32_t mutableProperties = element->mutableProperties();
+        if (mutableProperties & (WebCompositorMutablePropertyScrollTop | WebCompositorMutablePropertyScrollLeft))
+            style.setWillChangeScrollPosition(true);

[ajuma, 2015/11/20 21:55:27]

will-change:scroll-position doesn't currently create a composited layer (or do
anything really), so you'll need to change
ComputeStyle::hasWillChangeCompositedHint. Also, will-change doesn't directly
force a PaintLayer (unless the property being will-changed is one that creates a
stacking context, in which case the computed z-index gets set to 0 which forces
a PaintLayer), but I guess if the thing is really overflow:scroll you'll already
have one.

[Ian Vollick, 2015/11/23 20:25:54]

Ah, whoops. Yes, the composited layer _doesn't_ actually depend on will-chang.
That was handled in a previous CL
(https://code.google.com/p/chromium/codesearch#chromium/src/third_party/WebKit...).

[chrishtr, 2015/11/24 16:22:50]

Was Ali's comment addressed? Not sure how to read your response.

Also, doesn't setting values in this function cause a developer-visible side
effect (will-change: scroll-position in ComputedStyle) even if CompositorWorker
is not web-exposed?

Please also unittest this function.

[Ian Vollick, 2015/11/25 18:28:44]

I was trying to say that will-change does not play a role in getting proxied
elements promoted. I.e., will-change is not required for the code to work.

I'd wanted proxies to imply will change because having a proxy pointed at a
property truly does mean that you will-change it. It would be helpful for the
engine to know this in case it wanted to make optimizations.

That said, you're right that this behavior will be developer visible when we
implement snap and sticky in terms of compositor worker. Since this will-change
business is really just an optimization for when (and if) CW is web exposed,
I've removed it from this patch.

+        Vector<CSSPropertyID> properties = style.willChangeProperties();
+        if ((mutableProperties & WebCompositorMutablePropertyOpacity) && !properties.contains(CSSPropertyOpacity))
+            properties.append(CSSPropertyOpacity);
+        if ((mutableProperties & WebCompositorMutablePropertyTransform) && !properties.contains(CSSPropertyTransform))
+            properties.append(CSSPropertyTransform);
+        style.setWillChangeProperties(properties);
+    }
+
     // Make sure our z-index value is only applied if the object is positioned.
     if (style.position() == StaticPosition && !parentStyleForcesZIndexToCreateStackingContext(parentStyle))
         style.setHasAutoZIndex();
 
     // Auto z-index becomes 0 for the root element and transparent objects. This prevents
     // cases where objects that should be blended as a single unit end up with a non-transparent
     // object wedged in between them. Auto z-index also becomes 0 for objects that specify transforms/masks/reflections.
     if (style.hasAutoZIndex() && ((element && element->document().documentElement() == element)
         || style.hasOpacity()
         || style.hasTransformRelatedProperty()
@@ skipping 290 matching lines @@
         // We want to count vertical percentage paddings/margins on flex items because our current
         // behavior is different from the spec and we want to gather compatibility data.
         if (style.paddingBefore().hasPercent() || style.paddingAfter().hasPercent())
             UseCounter::count(document, UseCounter::FlexboxPercentagePaddingVertical);
         if (style.marginBefore().hasPercent() || style.marginAfter().hasPercent())
             UseCounter::count(document, UseCounter::FlexboxPercentageMarginVertical);
     }
 }
 
 }