I introduced (sorry!) a failing layout test (editing/selection/last-empty-inline.html) for Windows and Qt in Bug 29966. The change was: http://trac.webkit.org/changeset/51429 I put the failing test as the URL of this bug. This test drags the mouse from a point below the comma between "Hello" and "world!" and check if "world!" is selected. Before Bug 29966, "world!" wasn't selected even if we drag the mouse from the point below '!' after "world" to the comma between "Hello" and "world". The last-empty-inline was intended to check this behavior. I'll submit a patch which changes the test case so that it drags the mouse from the point below '!'. Note that it seems Qt lacks something to test editing features. So, I couldn't fix this test for now. By the way, I'm not sure if the original test case should pass as is. In Mac world, this should be definitely pass (clicking below the last line is considered as the click to the end of the last line). In Windows, I'm not sure what is the right behavior. With Firefox, "world!" is selected but nothing is selected with WebKit. WebKit ignores empty inline elements when it searches the clicked inline element even if the empty inline element is visible because of paddings. So, WebKit considers "Hello, world!" is the last line and the clicking below the last line is consider as the click to the middle of the last line in Windows. If someone think this is a bug, I'll file another bug. Hmm... my English isn't good to describe this kind of a bit complicated geometric stuff. Please feel free to say "hey, your English is terrible, elaborate again" if there are something unclear.
Created attachment 44023 [details] Patch v1
> Patch v1 Note that I confirmed this test fails before Bug 29966 on Mac and Windows. So, I think this test is still working as the regression test of Bug 29966.
style-queue ran check-webkit-style on attachment 44023 [details] without any errors.
Comment on attachment 44023 [details] Patch v1 I'm not sure this is OK, but I guess I'm sure enough to rubber-stamp the change. rs=me
Committed r51834: <http://trac.webkit.org/changeset/51834>