Now that bug 1134968 is fixed, passing JS::NullPtr() to NewObjectWithGivenTaggedProto will use the global, so we should be able to nix explicit parents from callers.

Bobby, could you stamp the dom & xpconnect bits of this?

Comment on attachment 8569488 [details] [diff] [review] part 3. Stop passing parents to js::NewProxyObject Review of attachment 8569488 [details] [diff] [review]: ----------------------------------------------------------------- ::: js/ipc/WrapperOwner.cpp @@ -1019,5 @@ > obj = NewProxyObject(cx, > &CPOWProxyHandler::singleton, > v, > nullptr, > - junkScope, lol ::: js/src/proxy/ScriptedIndirectProxyHandler.cpp @@ +435,5 @@ > RootedObject handler(cx, NonNullObject(cx, args[0])); > if (!handler) > return false; > + RootedObject proto(cx); > + proto = args.callee().global().getOrCreateFunctionPrototype(cx); Looks like this'll all fit in a single 99ch line now, probably?

> Looks like this'll all fit in a single 99ch line now, probably? Yep, done.

https://hg.mozilla.org/integration/mozilla-inbound/rev/d5e658bc0792 https://hg.mozilla.org/integration/mozilla-inbound/rev/a00f2530c911 https://hg.mozilla.org/integration/mozilla-inbound/rev/2faa0e91fe23 https://hg.mozilla.org/integration/mozilla-inbound/rev/331761aaae22

All four backed out in https://hg.mozilla.org/integration/mozilla-inbound/rev/7c3cc4f11d38 as I grasp at straws to figure out what broke mochitest-e10s-dt today: https://treeherder.mozilla.org/logviewer.html#?job_id=7050990&repo=mozilla-inbound

RElanded: https://hg.mozilla.org/integration/mozilla-inbound/rev/2dac197193b7 https://hg.mozilla.org/integration/mozilla-inbound/rev/84c7cadc7dd2 https://hg.mozilla.org/integration/mozilla-inbound/rev/2e9b1150861b https://hg.mozilla.org/integration/mozilla-inbound/rev/eafeefc2a038

https://hg.mozilla.org/mozilla-central/rev/2dac197193b7 https://hg.mozilla.org/mozilla-central/rev/84c7cadc7dd2 https://hg.mozilla.org/mozilla-central/rev/2e9b1150861b https://hg.mozilla.org/mozilla-central/rev/eafeefc2a038