Removed a use of markDOMObjectWrapper: Element and Document
Created attachment 90674 [details] Patch
Attachment 90674 [details] did not build on chromium: Build output: http://queues.webkit.org/results/8498245
Attachment 90674 [details] did not build on chromium: Build output: http://queues.webkit.org/results/8497280
Comment on attachment 90674 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=90674&action=review > LayoutTests/fast/dom/script-tests/dataset-gc.js:26 > var dataset = d.dataset; > +dataset.__proto__.customProperty = 1; // Add a property to our prototype, so it won't forward to our element. > +dataset.customProperty = 1; // Now set a property on ourselves. > +shouldBe("d.dataset.customProperty", "1"); > > -// Ensure there is a __proto__ before test. > -shouldBeNonNull("d.dataset.__proto__"); > - > -// Set __proto__ to non-starting value. (Must be null do to __proto__ restrictions). > -d.dataset.__proto__ = null; > -shouldBeNull("d.dataset.__proto__"); > - > -// Null out reference to the dataset. > dataset = null; > > gc(); > > -// Test that the null persisted the GC. > -shouldBeNull("d.dataset.__proto__"); > +// Test that the custom property persisted the GC. > +shouldBe("d.dataset.customProperty", "1"); If you are going to change this test to not use __proto__, you should ensure/assert that setting dataset.customProperty does not add an attribute to the underlying element by checking for the attribute explicitly.
Comment on attachment 90674 [details] Patch r=me, but fix the test as sam said, and dont break chrome
Committed r84665: <http://trac.webkit.org/changeset/84665>