Paolo notified me pagedown makes lazy loading go mad, I also see a bunch of exceptions generated by ensureVisibleElementsAreActive.

This looks like a simple coordinates problem, we have a rounded border on elements, richlistitems may have a top/bottom margin and likely there could be some subpixel rounding. I implemented this solution to verify that theory and so far I didn't see the exception, though it's maybe not the most elegant thing we may do.

I wonder if we should just try with NodesFromRect

I beilive so.

parseInt?

Comment on attachment 700337 [details] [diff] [review] patch v1.0 ok, let's try nodesFromRect and see if it's less bogus.

Comment on attachment 700576 [details] [diff] [review] patch v1.1 >+ // To ensure proper keyboard navigation we muse activate also the last >+ // non-visible node and the first non-visible node. I think this would read better: "Also activate the first invisible nodes in both boundaries (that is, above and below the visible area) to ensure proper keyboard navigation in both directions." >+ let nodeAfterVisibleArea = lastVisibleNode && lastVisibleNode.nextSibling; ... >+ let nodeBeforeVisibleArea = >+ firstVisibleNode && firstVisibleNode.previousSibling; after->below, before->above.

Comment on attachment 700609 [details] [diff] [review] patch v1.2 [Approval Request Comment] Bug caused by (feature/regressing bug #): Downloads panel feature User impact if declined: incomplete ui Testing completed (on m-c, etc.): m-i (merge pending) Risk to taking this patch (and alternatives if risky): limited to the feature String or UUID changes made by this patch: none

Verified as fixed on Firefox 20 on Windows 7, Ubuntu 12.04 and Mac OS X 10.8: Build ID: 20130326150557 Mozilla/5.0 (Windows NT 6.1; rv:20.0) Gecko/20100101 Firefox/20.0 Mozilla/5.0 (X11; Linux i686; rv:20.0) Gecko/20100101 Firefox/20.0 Mozilla/5.0 (Macintosh; Intel Mac OS X 10.8; rv:20.0) Gecko/20100101 Firefox/20.0