<html>
<head>
<base href="https://bugs.webkit.org/">
</head>
<body><span class="vcard"><a class="email" href="mailto:simon.fraser@apple.com" title="Simon Fraser (smfr) <simon.fraser@apple.com>"> <span class="fn">Simon Fraser (smfr)</span></a>
</span> changed
<a class="bz_bug_link
bz_status_NEW "
title="NEW - Position:fixed layers shouldn't allocate a backing buffer if all children are offscreen"
href="https://bugs.webkit.org/show_bug.cgi?id=235420">bug 235420</a>
<br>
<table border="1" cellspacing="0" cellpadding="8">
<tr>
<th>What</th>
<th>Removed</th>
<th>Added</th>
</tr>
<tr>
<td style="text-align:right;">Attachment #449611 is obsolete</td>
<td>1
</td>
<td>
</td>
</tr></table>
<p>
<div>
<b><a class="bz_bug_link
bz_status_NEW "
title="NEW - Position:fixed layers shouldn't allocate a backing buffer if all children are offscreen"
href="https://bugs.webkit.org/show_bug.cgi?id=235420#c3">Comment # 3</a>
on <a class="bz_bug_link
bz_status_NEW "
title="NEW - Position:fixed layers shouldn't allocate a backing buffer if all children are offscreen"
href="https://bugs.webkit.org/show_bug.cgi?id=235420">bug 235420</a>
from <span class="vcard"><a class="email" href="mailto:simon.fraser@apple.com" title="Simon Fraser (smfr) <simon.fraser@apple.com>"> <span class="fn">Simon Fraser (smfr)</span></a>
</span></b>
<pre>Comment on <span class=""><a href="attachment.cgi?id=449611&action=diff" name="attach_449611" title="Patch">attachment 449611</a> <a href="attachment.cgi?id=449611&action=edit" title="Patch">[details]</a></span>
Patch
View in context: <a href="https://bugs.webkit.org/attachment.cgi?id=449611&action=review">https://bugs.webkit.org/attachment.cgi?id=449611&action=review</a>
<span class="quote">> Source/WebCore/rendering/RenderLayerBacking.cpp:2793
> +static bool intersectsWithAncestor(const RenderLayer& ancestor, const LayoutRect& ancestorCompositedBounds, const RenderLayer& child)</span >
I would put child as the first argument.
<span class="quote">> Source/WebCore/rendering/RenderLayerBacking.cpp:2798
> + return true;</span >
It would be more truthful to have the function return std::optional<bool> and return nullopt here.
<span class="quote">> Source/WebCore/rendering/RenderLayerBacking.cpp:2802
> + LayoutPoint delta = child.convertToLayerCoords(&ancestor, LayoutPoint(), RenderLayer::AdjustForColumns);</span >
auto delta. Maybe also "{ }" instead of LayoutPoint().
It's unfortunate that this function is called inside a descendant layer tree walk, then your loop above, and child.convertToLayerCoords() both do ancestor tree walks, which can be a perf issue (we've seen this in the past). Maybe traverseVisibleNonCompositedDescendantLayers() can accumulate the "offset from ancestor" as it's doing the descendant traversal.
Probably fine in this case (we don't expect a lot of ancestor walking) but generally to be avoided.
<span class="quote">> Source/WebCore/rendering/RenderLayerBacking.cpp:2803
> + LayoutRect overlap = child.overlapBounds();</span >
auto
<span class="quote">> Source/WebCore/rendering/RenderLayerBacking.cpp:2804
> + overlap.move(delta.x(), delta.y());</span >
overlap.moveBy(delta);
<span class="quote">> LayoutTests/compositing/backing/no-backing-for-offscreen-children-of-position-fixed.html:19
> + <div style="position: fixed;z-index: -1;left: 0;top: 0;right: 0;bottom: 0;opacity: 0.5;"></span >
I don't think the z-index: -1 is relevant. Also this layer doesn't need to be full height or width (and making it so makes it harder to identify in the layer output). Also the opacity is not required.</pre>
</div>
</p>
<hr>
<span>You are receiving this mail because:</span>
<ul>
<li>You are the assignee for the bug.</li>
</ul>
</body>
</html>