#1708

Note that the Travis build failed since it wasn't run through grunt before the pull request was made. Also there are no unit tests. I believe there are already tests for the offset() method to check these cases, but if we're not handling them we would also need some corresponding unit tests. For more help go to http://contribute.jquery.org.

I ran "npm run build" I just didn't notice the errors.

I have added some tests.

Sorry about the delay, I meant to summon @mikesherov on this to be sure the interpretation was correct.

@mikesherov ping

@jquery/core Does anyone have an issue with landing this?

I'm interested in the size difference between two assignment operations on each parentOffset property vs. one (e.g., parentOffset.top += jQuery.css(…) - offsetParent.scrollTop()), but the logic and test changes look good to me.

I'm interested in the size difference between two assignment operations on each parentOffset property vs. one (e.g., parentOffset.top += jQuery.css(…) - offsetParent.scrollTop())

It saves 3 bytes.

Landing with the proposed change from Richard.

This PR caused a couple of problems in jQuery 2.2.0 & 1.12.0 so we've reverted it in 49833f7; we'll try to tackle the problem later.