cross-document is definitely a separate thing, and hopefully not something
we need to think about in the context of shadow DOM.

(In reply to Olli Pettay from comment #1)
> cross-document is definitely a separate thing, and hopefully not something
> we need to think about in the context of shadow DOM.

+1 though I'm not sure where I stand on the cross-origin web component (with separate script environment). This might be part of the "deep tree" but should involve an entirely separate event object/ nested dispatch, I think.

I don't think deepTree will help us to spec the closed mode because a open child shadow tree of the closed shadow tree should be "considered" *closed* from an ancestor node tree.

I have an idea to explain the behavior of each APIs in term of the "reachability (visibility)" between two node trees.

Let me update this thread when I'm ready to explain.

Anne, Olli,

I've introduced the "an unclosed {tree/node} of" relation between two {node trees/nodes} so that closed shadow trees behave correctly as expected.

- http://w3c.github.io/webcomponents/spec/shadow/#dfn-unclosed-tree
- http://w3c.github.io/webcomponents/spec/shadow/#dfn-unclosed-node

See
- https://github.com/w3c/webcomponents/commit/503e27c6b85ceb5c1d270039c69fec529c079117

This is an example of usages:

- https://github.com/w3c/webcomponents/commit/644a0fa4a22589d15fd123fa98436c651a0d3a62


I appreciate if you could check whether these definitions meet your expectations or not.

An "unclosed" is just a tentative name.
The other candidates are "reachable", "visible" or anything you want to call.

Anyway, these terms are not exposed to APIs.

Thanks for some comments on github.
Let me close this bug. I don't think we need a "deepTree" terminology.