Steps to reproduce: 1. Download attached main.html and subframe.html. 2. Open main.html is a debug Safari build. 3. Close it. Results: an alert says that a WebFrame leaked.
<rdar://problem/6833859>
Created attachment 29916 [details] main document
Created attachment 29917 [details] subframe
Created attachment 29918 [details] patch
Comment on attachment 29918 [details] patch As with all FrameLoader voodoo, I'm not 100% sure on this one - but it seems quite safe and reasonable, and fixes the bug. I think it's a shame the FrameLoadDelegate dumping that DRT does doesn't make an effort to dump the error for didFailLoadWithError, but that enhancement is outside the scope of this patch. r+
http://trac.webkit.org/changeset/43106
This was rolled out in r43107, because of layout tests failing on build bots: <http://build.webkit.org/results/Tiger%20Intel%20Release/r43106%20(711)/results.html> <http://build.webkit.org/results/Leopard%20Intel%20Release%20(Tests)/r43106%20(795)/results.html> <http://build.webkit.org/results/Windows%20Release%20(Tests)/r43106%20(603)/results.html>
Comment on attachment 29918 [details] patch Clearing review flag since this caused regression test failures.
http://trac.webkit.org/changeset/43148