User Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10_11_6) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/51.0.2704.106 Safari/537.36 Steps to reproduce: a containing element, with the CSS will-change property set to 'transform', steals the offsetParent from any parent divs with a positional value ('relative/absolute') set. Actual results: see above Expected results: the parent with the position value set should have been defined as the offset parent, rather than the child div with 'will-change'..
Can you provide a standalone testcase?
Created attachment 8781142 [details] testcase without will-change: transform I confirm the behavior difference with those 2 testcases
It is a bug of CSSOM, and the behavior is defined in CSSOM View spec . And I think the behavior defined there doesn't make sense. offsetParent should return the first stacking context ancestor rather than just basing on the computed value of position property.  https://drafts.csswg.org/cssom-view/#dom-htmlelement-offsetparent
https://lists.w3.org/Archives/Public/www-style/2016Sep/0000.html " Algorithm of `Element.offsetParent` ----------------------------------- - RESOLVED: Adopt the Gecko/Edge behavior and specify that .offsetParent is based on the nearest abspos containing block or table cell " So it seems to me this bug is invalid.
Actually I usually prefer closing this kind of bug when I see the change to spec... But given the working group has resolved on that, I think we can close it now.