The test fast/dom/HTMLAnchorElement/anchor-nodownload-set.html tries to trigger a download for a Blob object. Our download code assumes all ResourceHandles have an associated FilterStream object (set on the ::start method for ResourceHandle), but Blob objects have their own ResourceHandle class which does not this (since they are not really retrieved from the network). This makes us crash when we ASSERT the stream object, since it will be null. As a workaround, call the right methods in our networking layer from FrameLoaderClient itself. Note that this does not yet fix this test, since there's at least another bug involved.
Created attachment 198507 [details] no-stream.diff
Created attachment 198508 [details] no-stream.diff Applies properly against trunk.
Comment on attachment 198508 [details] no-stream.diff View in context: https://bugs.webkit.org/attachment.cgi?id=198508&action=review > Source/WebKit/blackberry/ChangeLog:7 > + You can add "Internally reviewed by Joe Mason" to this changelog message. (Plus you should add the PR number if we have an internal PR for it.)
Ping about this? As Joe says, it's been already reviewed by him internally.
Comment on attachment 198508 [details] no-stream.diff Clearing flags on attachment: 198508 Committed r148880: <http://trac.webkit.org/changeset/148880>
All reviewed patches have been landed. Closing bug.