When a live iframe is moved to a different document via adoptNode, the frame's resources are still being tracked by the former page's ProgressTracker. When a resource load is transferred to a different document/page, the ProgressTracker should treat that item as being completed. (Decrementing the progress of bytes received seems wrong.) Need more study to determine how the new page's ProgressTracker is affected by the resource load being transferred to the new page.
Created attachment 72769 [details] patch
Attachment 72769 [details] did not build on qt: Build output: http://queues.webkit.org/results/5047005
Created attachment 72848 [details] updated patch - fix compile error If someone knows of a way to test this from JS, I'm all ears.
Comment on attachment 72848 [details] updated patch - fix compile error View in context: https://bugs.webkit.org/attachment.cgi?id=72848&action=review Looks good! One small iteration... > WebCore/loader/ResourceLoadNotifier.cpp:167 > + oldPage->progress()->completeProgress(identifier); Could we add ASSERT documenting that new page and old page are guaranteed different here? This code actually uses this assumptions (because otherwise it immediately deletes the just-transfered 'identifier' from the progress tracker).
Created attachment 72987 [details] updated patch - added assert
Comment on attachment 72987 [details] updated patch - added assert r=me
The commit-queue encountered the following flaky tests while processing attachment 72987 [details]: animations/suspend-resume-animation.html compositing/video/video-background-color.html Please file bugs against the tests. These tests were authored by cmarrin@apple.com and simon.fraser@apple.com. The commit-queue is continuing to process your patch.
Comment on attachment 72987 [details] updated patch - added assert Clearing flags on attachment: 72987 Committed r71625: <http://trac.webkit.org/changeset/71625>
All reviewed patches have been landed. Closing bug.