Bug 93753 – Nested shadow root distributes node to insertion point that is not its child

• Component: DOM
• Status: RESOLVED INVALID

[Steve Orvell]

Created attachment 157835 [details]
Reduction showing actual and expected output.

1. construct a shadowRoot on an element with a selection point that distributes inside another element (B) in the shadowRoot.
2. on B, construct a shadowRoot with a selection point which distributes all content (e.g. <content></content>)

Children of the host containing B are selected into it. This appears to violate Section 5.3 number 1 of the spec (https://dvcs.w3.org/hg/webcomponents/raw-file/tip/spec/shadow/index.html#matching-insertion-points).

Note however, the spec may change on this point: https://www.w3.org/Bugs/Public/show_bug.cgi?id=18513

[Dominic Cooney]

I think the implemented behavior is correct, per Section 5.7 [1], step 2. (second bullet). 1. 2 "If CHILD is itself an insertion point, and the tree being rendered is nested, render them using insertion point rendering steps."

You mentioned Section 5.3, I think you are confusing selection and rendering. First consider selection:

(A) <content select="*"></content>

is selecting

(B) <content select=".foo"></content>

That in turn is selecting

(C) <div class="foo">I am a foo.</div>

Next consider rendering:

With reference to Section 5.7, when step 2. (second bullet). 1 is processing (A), CHILD is B.

At step 2. (second bullet). 1. 2. the insertion point rendering recurses to process (B) (ie back to step 2. (second bullet.)).

At step 2. (second bullet). 1. 1. (C) is rendered.

[1] <https://dvcs.w3.org/hg/webcomponents/raw-file/7a24eae26c59/spec/shadow/index.html#rendering-shadow-subtrees>

[Dimitri Glazkov (Google)]

Dominic is right! I completely missed the point that the actual <content> element is the one being selected into the insertion point.