22420 2013-06-20 20:30:21 +0000 PopStateEvent.state should be a copy (via structured clone) of the value it was initialized to 2013-07-02 22:33:09 +0000 1 1 1 Unclassified WHATWG HTML unspecified Other other RESOLVED WONTFIX http://www.whatwg.org/specs/web-apps/current-work/#dom-popstateevent-state P3 normal Unsorted 1 contributor ian adamk ian mike contributor oldest_to_newest 89660 0 contributor 2013-06-20 20:30:21 +0000 Specification: http://www.whatwg.org/specs/web-apps/current-work/multipage/history.html Multipage: http://www.whatwg.org/C#dom-popstateevent-state Complete: http://www.whatwg.org/c#dom-popstateevent-state Referrer: Comment: PopStateEvent.state should be a copy (via structured clone) of the value it was initialized to, not the value itself. This would make events generated by the user agent and those generated by code calling new PopStateEvent() behave the same way. Posted from: 216.239.45.83 User agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10_8_4) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/29.0.1543.2 Safari/537.36 89661 1 adamk 2013-06-20 20:37:58 +0000 This concern looks like it should also apply to MessageEvent, which uses the same wording and has the same difference between events created via new MessageEvent and those created via, e.g., postMessage. 89663 2 ian 2013-06-20 21:00:33 +0000 I don't understand. Why would we want to prevent an author from creating an event with the specific data they set? That would be quite a radical departure from the behaviour of other event objects. 89664 3 adamk 2013-06-20 21:08:28 +0000 I guess my argument is for consistency of behavior. Why should it be possible to create a MessageEvent with an HTMLDocument as its data when using the constructor form but not when using postMessage? I take it your response is that, in the constructor case, there's no need for restricting authors to those things which are structured-cloneable. My motivation in filing this bug happens to involve implementation details of these events in Blink, where storing the passed-in object itself creates an opportunity for memory leaks and security bugs. But there are alternative fixes for those problems if it turns out that this inconsistency between organic and synthetic events is intentional. 90165 4 ian 2013-07-02 22:33:09 +0000 Closing per IRC discussion last week; reopen if I misremember our conclusion.