[webkit-reviews] review granted: [Bug 24624] Crash in imageLoadEventTimerFired after adoptNode used on <img>, seen with inspector, which uses adoptNode : [Attachment 28656] patch, this time with a regression test

Tue Mar 17 01:46:46 PDT 2009

Alexey Proskuryakov <ap at webkit.org> has granted Darin Adler <darin at apple.com>'s
request for review:
Bug 24624: Crash in imageLoadEventTimerFired after adoptNode used on <img>,
seen with inspector, which uses adoptNode
https://bugs.webkit.org/show_bug.cgi?id=24624

Attachment 28656: patch, this time with a regression test
https://bugs.webkit.org/attachment.cgi?id=28656&action=review

------- Additional Comments from Alexey Proskuryakov <ap at webkit.org>
+	 member functions. Got rid of pointless boolean bit fields that don't
make the
+	 object any smaller, since 3 bits and 3 bytes both cost 4 bytes.

On PowerPC, bool is still 4 bytes - I don't think that we should regress PPC,
especially when that's easy to avoid.

+    m_dispatchingList = m_dispatchSoonList;
+    m_dispatchSoonList.clear();

A swap would be more efficient than copying.

Index: LayoutTests/fast/dom/HTMLImageElement/resources/image-load-subframe.html

===================================================================

--- LayoutTests/fast/dom/HTMLImageElement/resources/image-load-subframe.html   
(revision 0)
+++ LayoutTests/fast/dom/HTMLImageElement/resources/image-load-subframe.html   
(revision 0)
@@ -0,0 +1 @@
+<body><img src="blue_rect.jpg"></body>

With a data: URL, the test should be easier to read. It isn't great that a
failure in this test is detected as a crash in another one.

r=me
