Bugzilla

Here is another similar case which we aren't handling correctly currently: if the popover invoker has children, then they are never visited in sequential focus navigation. Well maybe the right solution here is that when we get to a popover invoker, we should visit the popover, then visit its children or shadow root. That seems to be what Chrome does. (The spec says we should use shadow-including tree order to decide which to visit first, but I don't think the position of the popover in the tree should affect its position in focus navigation. Maybe worth filing a spec issue about that.)

Here is another similar case which we aren't handling correctly currently: if the popover invoker has children, then they are never visited in sequential focus navigation. Well maybe the right solution here is that when we get to a popover invoker, we should visit the popover, then visit its children or shadow root. That seems to be approximately what Chrome does, although they visit the shadow root before the popover… (The spec says we should use shadow-including tree order to decide which to visit first, but I don't think the position of the popover in the tree should affect its position in focus navigation. Maybe worth filing a spec issue about that.)

Here is another similar case which we aren't handling correctly currently: if the popover invoker has children, then they are never visited in sequential focus navigation. Well maybe the right solution here is that when we get to a popover invoker, we should visit the popover, then visit its children or shadow root. That seems to be approximately what Chrome does, although they visit the shadow root before the popover… (The spec says we should use shadow-including tree order to decide which to visit first, but I don't think the position of the popover in the tree should affect its position in focus navigation. Maybe worth filing a spec issue about that: https://github.com/whatwg/html/issues/12871.)