<html>
<head>
<base href="https://bugs.webkit.org/" />
</head>
<body>
<p>
<div>
<b><a class="bz_bug_link
bz_status_NEW "
title="NEW - REGRESSION (r206247): Painting milestones can be delayed until the next layer flush"
href="https://bugs.webkit.org/show_bug.cgi?id=164340#c8">Comment # 8</a>
on <a class="bz_bug_link
bz_status_NEW "
title="NEW - REGRESSION (r206247): Painting milestones can be delayed until the next layer flush"
href="https://bugs.webkit.org/show_bug.cgi?id=164340">bug 164340</a>
from <span class="vcard"><a class="email" href="mailto:mitz@webkit.org" title="mitz@webkit.org <mitz@webkit.org>"> <span class="fn">mitz@webkit.org</span></a>
</span></b>
<pre>Comment on <span class=""><a href="attachment.cgi?id=293734&action=diff" name="attach_293734" title="Have WebKit, not WebCore, defer sending the milestones until after the commit">attachment 293734</a> <a href="attachment.cgi?id=293734&action=edit" title="Have WebKit, not WebCore, defer sending the milestones until after the commit">[details]</a></span>
Have WebKit, not WebCore, defer sending the milestones until after the commit
View in context: <a href="https://bugs.webkit.org/attachment.cgi?id=293734&action=review">https://bugs.webkit.org/attachment.cgi?id=293734&action=review</a>
<span class="quote">>> Source/WebCore/ChangeLog:11
>> + the call to firePaintRelatedMilestonesIfNeeded until after the commit.
>
> Was it deferred for a good reason, like coalescing?</span >
It was deferred in order to make sure that WebKit client gets notified only after the commit (see <a class="bz_bug_link
bz_status_RESOLVED bz_closed"
title="RESOLVED FIXED - Still possible to fire paint-related LayoutMilestones before CA has committed the changes"
href="show_bug.cgi?id=115753">bug 115753</a>). That bug was probably reintroduced when <<a href="https://trac.webkit.org/r166015">https://trac.webkit.org/r166015</a>> (the fix for <a class="bz_bug_link
bz_status_RESOLVED bz_closed"
title="RESOLVED FIXED - FrameView::paintContents() is not called for composited content"
href="show_bug.cgi?id=130541">bug 130541</a>) made RenderLayerBacking::paintIntoLayer() call FrameView::didPaintContents(), which fires the paint-related milestones synchronously. That is, until <<a href="https://trac.webkit.org/r206247">https://trac.webkit.org/r206247</a>> introduced a completely different way for the milestones to be deferred, which happened to defer them too much in some cases.
<span class="quote">>> Source/WebKit2/WebProcess/WebPage/mac/TiledCoreAnimationDrawingArea.mm:428
>> + if (TiledCoreAnimationDrawingArea* drawingArea = static_cast<TiledCoreAnimationDrawingArea*>(retainedPage->drawingArea())) {
>
> auto foo = downcast<TiledCoreAnimationDrawingArea>() ?
>
> When is drawingArea != this? Very confusing to fetch it here again.</span >
drawingArea is null if in the mean while WebPage::close has been called.
<span class="quote">>> Source/WebKit2/WebProcess/WebPage/mac/TiledCoreAnimationDrawingArea.mm:439
>> + drawingArea->m_pendingNewlyReachedLayoutMilestones = 0;
>
> ...which you could also call here.</span >
I waffled on this. It will be a similar number of lines of code, and once 10.10 support is gone, it will be even more silly for it to be a function.</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>