[webkit-reviews] review granted: [Bug 204219] JS wrappers of scroll event targets can get prematurely collected by GC : [Attachment 383603] Fixes the bug

[bugzilla-daemon at webkit.org]

Keith Miller <keith_miller at apple.com> has granted Ryosuke Niwa
<rniwa at webkit.org>'s request for review:
Bug 204219: JS wrappers of scroll event targets can get prematurely collected
by GC
https://bugs.webkit.org/show_bug.cgi?id=204219

Attachment 383603: Fixes the bug

https://bugs.webkit.org/attachment.cgi?id=383603&action=review




--- Comment #6 from Keith Miller <keith_miller at apple.com> ---
Comment on attachment 383603
  --> https://bugs.webkit.org/attachment.cgi?id=383603
Fixes the bug

View in context: https://bugs.webkit.org/attachment.cgi?id=383603&action=review

r=me.

>>> Source/WebCore/dom/Document.cpp:373
>>> +};
>> 
>> I thought you said there was a pending activity class? Can we just use that
instead?
> 
> No, we can't make every ContainerNode an active DOM object because
ContainerNode is a very common object in DOM.

Oh, I thought the system worked with any node, presumably via GCReachableRef.

> Source/WebCore/dom/Document.cpp:4021
> +    if (targets.findMatching([&target] (auto& entry) { return entry.ptr() ==
⌖ }) != notFound)

Why not just [&]? It's all gonna get inlined anyway.