1496510 - Turn on the behavior from bug 1329324 (disallowing defining non-configurable properties on a WindowProxy) on beta/release

Status: RESOLVED WONTFIX

(Core :: DOM: Bindings (WebIDL), defect, P2)

Description

[Boris Zbarsky [:bzbarsky]]

+++ This bug was initially created as a clone of Bug #1178638 +++

Need to see how web-compatible this is first...

Comment 1

[Andrew McCreight [:mccr8]]

As seen in Bug 1699758, there is now a WPT for some of this behavior, so we're not behaving consistently between Nightly and release on the test.

Comment 2

[Tom S [:evilpie]]

Do we have any plans for this going forward? Is there any discussion besides https://github.com/tc39/ecma262/pull/688 ?

Comment 3

[Mark S. Miller]

As I recall, incompat with one library blocked the simpler solution that Mozilla originally implemented. IIRC that library has since been fixed. If so, perhaps we can return to that simpler solution. This would be great progress.

The status quo does have a significant hazard: the enforcement of the object invariants by proxies is inductive. It relies on there being no possible target on which the object invariants can be violated. Introduce such a violating object that proxies can use as a target, and you enable proxies to violate the invariants as well.

What we need are test262 tests that fail on the status quo because of the presence of an object that violates these invariants. That plus a less painful way of satisfying these invariants (the original Mozilla implementation) should bring the engines into alignment. No one wants a persistent test262 violation in their public test results.

Comment 4

[Anne (:annevk)]

Tom, I think that is all the discussion. The status quo is the HTML Standard overriding ECMAScript (see "willful violation").

If the Nightly behavior is problematic we could align with Release for now, until it's sorted on the TC39 side.

Mark, I don't really know what bandwidth we have to run further experiments here. It would be great if Chrome/Safari could pursue those first.

(Moving this to Bindings as that seems like a suitable place to track this rather than the catch-all component.)

Comment 5

[Anne (:annevk)]

Another test was added to WPT by Chromium which resulted in bug 1732128. Removing this on Nightly seems even more reasonable as there's still no plan here or people to work on it.

Comment 6

[Anne (:annevk)]

Edgar, Peter, given that this now results in Google Analytics breakage could either of you please align Nightly with Release? This isn't worth our time given that Chrome/Safari are not actively pursuing a fix.

Comment 7

[Edgar Chen [:edgar]]

Okay, so lets wontfix this, I will align Nightly with Release in bug 1758164.