Since new results were landed in r57109, editing/pasteboard/drag-image-to-contenteditable-in-iframe.html has been failing on Windows.
<rdar://problem/7832412>
I need the help from someone with a Windows machine who can regenerate the results of this test.
(In reply to comment #2) > I need the help from someone with a Windows machine who can regenerate the > results of this test. Darin, it looks to me like the Windows bots are generating the same results that the Mac bots generated prior to r57109. i.e., the results coming out of the Windows bots are identical to the expected results from r57108. Does this indicate that the bug is not fixed on Windows?
Alexey says I can just get the results from the bots. I will do that now.
(In reply to comment #3) > Darin, it looks to me like the Windows bots are generating the same results > that the Mac bots generated prior to r57109. i.e., the results coming out of > the Windows bots are identical to the expected results from r57108. Does this > indicate that the bug is not fixed on Windows? That would be exceedingly surprising. There is no platform-specific element involved here. I am not sure how to figure it out.
(In reply to comment #5) > (In reply to comment #3) > > Darin, it looks to me like the Windows bots are generating the same results > > that the Mac bots generated prior to r57109. i.e., the results coming out of > > the Windows bots are identical to the expected results from r57108. Does this > > indicate that the bug is not fixed on Windows? > > That would be exceedingly surprising. There is no platform-specific element > involved here. I am not sure how to figure it out. But I see that you are correct. This is a mystery.
I have now checked and the bug that r57109 is indeed fixed on Windows as well as Mac. But this dragging test works differently because of a difference in the code path on Windows. On Windows, the new <img> element has a full path to the image. On Mac, the new <img> element has the same src attribute as the original <img> element. So it's not surprising that this test has different results, and it has nothing to do with the bug I fixed, but it does represent a platform difference.
http://trac.webkit.org/changeset/57160
I filed bug 37162 about the problem.