Progress on WebKit this week happens in various areas.
On the testsuite first: I fixed several small issues that triggered asserts when WebKit was built with asserts enabled. This includes a problem with the sequencing of events when loading an invalid URL, and a double deletion of an object when iframes are involved. These two problems could have created some real-world issues, so WebKit should be a little bit more stable now. Another problem was the lack of "key up" events and mixup of keycodes vs characters in the testsuite keyboard simulator, which prevented us to test the editing code in an useful way. Another problem was some browser settings were modified by some tests (such as the text size, and page zoom factor), and not reset before running the following tests. This led to some unexpected errors which are now avoided. With these issues fixed, I can have a look at the remaining failing tests, knowing that they are more likely to uncover actual bugs.