The test at http://www.w3schools.com/js/tryit.asp?filename=tryjs_anchor1 triggers a crash inside the loader. I have a test case and fix underway.
Created attachment 28359 [details] patch
Comment on attachment 28359 [details] patch >+shouldBe("testWindow.location.toString()", "'/'"); // Firefox returns about:blank >+shouldBe("testWindow.location.href", "'/'"); // Firefox returns about:blank >+shouldBe("testWindow.location.protocol", "':'"); // Firefox returns about: >+shouldBe("testWindow.location.host", "''"); // Firefox throws an exception >+shouldBe("testWindow.location.hostname", "''"); // Firefox throws an exception This patch looks good to me. (I verified that all of the URL strings have been completed before the isEmpty checks.) However, I wonder about all of these behavior differences from Firefox. Are those potential compat issues?
(In reply to comment #2) > This patch looks good to me. (I verified that all of the URL strings have > been completed before the isEmpty checks.) However, I wonder about all of > these behavior differences from Firefox. Are those potential compat issues? They probably aren't but they might be. That's why I added those comments. We should check the behavior of IE before making any changes. I'd definitely be open to changing this to be closer to the other browsers. And we should check if HTML 5 has anything to say about this.
> They probably aren't but they might be. That's why I added those comments. See bug 21597, currently in review queue.
Although I landed this patch, I just noticed that while the regression test works in Safari, in DumpRenderTree it's just FAIL, FAIL, FAIL, so I need to resolve that.
Comment on attachment 28359 [details] patch Clearing review flag since this was landed.
http://trac.webkit.org/changeset/41484 http://trac.webkit.org/changeset/41502