Test that WindowProxy's [[Set]] fails early for OOB indices #29377
Conversation
wpt-pr-bot
requested review from
annevk,
domenic,
foolip,
jdm,
jgraham and
zqzhang
| window[2] = 2; | ||
| assert_false(window.hasOwnProperty(2)); | ||
| Object.defineProperty(EventTarget.prototype, 42, { set: t.unreached_func("EventTarget.prototype[42] setter should be unreachable!") }); | ||
| assert_false(Reflect.set(window, 42, "foo")); |
There was a problem hiding this comment.
Is there a particular reason to use Reflect.set here?
There was a problem hiding this comment.
In sloppy mode, there is no way to test return value of [[Set]] being false except using Reflect.set.
Chrome returns true instead of false.
There was a problem hiding this comment.
Can you use a normal set here, and put the Reflect check in a separate test() below?
|
Hmm maybe I shouldn't have merged the spec PR so quickly. You claim that all browsers have the behavior, but https://wpt.fyi/results/html/browsers/the-window-object/window-indexed-properties.html?diff&filter=ADC&run_id=5103827915112448&run_id=5652311210721280 shows that Chrome does not? |
|
@domenic Chrome fails newly added test for the same reason it fails the existing |
|
I mean, if Reflect.set doesn't return false, then how do you know it failed to set? |
|
Since |
Ref. whatwg/html#6767.