prefable: True has been removed by bug 814821.

https://tbpl.mozilla.org/?tree=Try&rev=4c3066547131 Hm, this patch caused the following regressions. Bug 763341: new XMLHttpRequest().QueryInterface(Components.interfaces.nsIClassInfo) fails (but new XMLHttpRequest().QueryInterface(Components.interfaces.nsIXMLHttpRequest) succeeds) Bug 317448, Bug 435425: new XMLHttpRequest() instanceof Components.interfaces.nsIXMLHttpRequest fails (but new XMLHttpRequest() instanceof XMLHttpRequest succeeds) Is this expected?

> new XMLHttpRequest().QueryInterface(Components.interfaces.nsIClassInfo) fails That seems fine to me. > new XMLHttpRequest() instanceof Components.interfaces.nsIXMLHttpRequest fails Ah. That's because nsJSIID::HasInstance checks for classinfo.... Peter, would it make sense to just fall back to QI there if no classinfo on a new-binding object?

This part should have no side effect. https://tbpl.mozilla.org/?tree=Try&rev=9c4f69ab853b

Comment on attachment 688757 [details] [diff] [review] Part 1: Remove legacy XHR constructor r=me

(I don't know what info was needed from me) I assume QIing in general doesn't fail.

- test_wrappers.xul will need an update again when Screen is unpref'ed. But I think we should have a test until the last classinfo is removed from new-bindings objects. - test_bug317448.html has been removed because it doesn't make sense in the new-bindings world. - test_bug435425.html is now using regular constructor objects instead of SpecialPowers.Ci.nsIFoo. Try result: https://tbpl.mozilla.org/?tree=Try&rev=ff1cb49b1ff8

Comment on attachment 690087 [details] [diff] [review] Part 2: Remove DOMClassInfo remnants from XMLHttpRequest r=me. Thank you!