[Regression] Chrome 52 shows non enumerable variables on proxy objects. 51 was ok
Reported by
denis.mi...@gmail.com,
Jul 28 2016
|
|||||||||
Issue descriptionUserAgent: Mozilla/5.0 (Windows NT 10.0; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/52.0.2743.82 Safari/537.36 Steps to reproduce the problem: 1. Open the attacched file into chrome 52 and look at console logs. A message will show up if test pass or not 2. All arrays must be empty to be ok What is the expected behavior? a proxy object should not return non enumerable properties when Object.keys() is called. What went wrong? a proxy object return also non enumerable properties when Object.keys() is called. Did this work before? Yes Chrome 51 was ok. Issue begin on Chrome 52 Chrome version: 52.0.2743.82 Channel: stable OS Version: 10.0 Flash Version: Shockwave Flash 22.0 r0 I've got 4 applications stop working due to this. It's very serius to me.
,
Jul 28 2016
This has been fixed in V8 (see https://bugs.chromium.org/p/v8/issues/detail?id=5174) and merged back to 5.3 but not (yet?) 5.2.
,
Jul 28 2016
,
Jul 28 2016
,
Jul 28 2016
,
Jul 28 2016
,
Jul 28 2016
The fix depends on 4-5 (or more) CLs which occurred since 5.2, and may be too much to merge back to 5.2. I've asked cbruni's opinion on the V8 bug
,
Jul 28 2016
I feel uncomfortable backmerging the 4-ish CLs. The original cause for this issue was the discrepancy in the spec between how Object.keys and how for-in treated enumerable keys of proxies. Compared to the newer Key-Accumulator it's rather tricky to get it right in V8 5.2. This is fixed though in the current beta M53. What I will try is to make M52 behave like M51, aka not breaking the Object.keys case.
,
Jul 29 2016
,
Jul 29 2016
,
Jul 29 2016
Can you add also a unit test to avoid this regression in next releases?
,
Jul 29 2016
See comment 2, we've detected this issue and added already a test accordingly. CL to fix the issue in M52 is on the way.
,
Aug 28 2016
Are there an estimated release date for this? This is a show stopper for me and I've to know what to do to fix (if wait for chrome fix or re-engeneering on my side). Tks
,
Aug 30 2016
Sorry, I was on vacation. The original fix was ready but there is now no time to make it into M52 as the next release (M53 which is fixed) is going to happen within a week from now. Hence I have to mark this as WontFix as there is nothing we can do at this point. |
|||||||||
►
Sign in to add a comment |
|||||||||
Comment 1 by neis@chromium.org
, Jul 28 2016