Intent to Implement and Ship: Node.getRootNode(options)

[Hayato Ito]

Contact emails

hay...@chromium.org

Spec

https://dom.spec.whatwg.org/#dom-node-getrootnode

Summary

Node.getRootNode() is the new name for Node.rootNode [1].

Blink had shipped Node.rootNode, however, it turned out that it was not Web compatible [2], and Node.rootNode was removed from Blink [3].

Node.getRootNode() is the new name on which browser vendors agreed as the new name for Node.rootNode.

[1]: Intent to Ship: Node.rootNode

[2]: Node.prototype.rootNode is not Web compatible #241

[3]: Revert of Ship Node.rootNode

Interoperability and Compatibility Risk

This is a new feature, hoping this name is Web compatible.

Ongoing technical constraints

No.

Will this feature be supported on all six Blink platforms (Windows, Mac, Linux, Chrome OS, Android, and Android WebView)?

Yes.

OWP launch tracking bug

https://bugs.chromium.org/p/chromium/issues/detail?id=629739

Link to entry on the feature dashboard

https://www.chromestatus.com/feature/5736827696709632

Requesting approval to ship?

Yes.

[Elliott Sprehn]

This isn't just a rename, it's adding features at this same time with the options dictionary. Hopefully in the future we don't add features at the same time we fix name conflicts. :)

It's really hard to parse that spec language, in examples what do the two modes do?

[Hayato Ito]

Yeah, that's a good point. Sorry for not mentioning it.

An optional parameter allows us to get the root's host's root's host's root's host's root, .... recursively, until it reaches the most *ancestor* root.

Example:

let host1 = document.createElement('div');

let root1 = host1.attachShadow({ mode: 'open' };

let host2 = document.createElement('div');

root1.appendChild(host2);

let root2 = host2.attachShadow({ mode: 'open' };

let child2 = root2.appendChild

root2.appendChild(child2)

Now, we have the following hierarchy:

<div> # host1

  <#root1>   # host1's shadow root

    <div> #host2

      <#root2> # host2's shadow root

        <div>  # child2

Optional parameter changes the getRootNodes(optional)'s behavior as follows:

assert(child2.getRootNode() == root2);

assert(child2.getRootNode({ composed: true}) == host1);

assert(host2.getRootNode() == root1);

assert(host2.getRootNode({ composed: true}) == host1);

assert(host1.getRootNode() == host1);

assert(host1.getRootNode({ composed: true}) == host1);

If we append host1 to document.body:

> document.body.appendChild(host1)

Then, getRootNode({ composed: true}) would return a document.

assert(child2.getRootNode() == root2);

assert(child2.getRootNode({ composed: true}) == document);

assert(host2.getRootNode() == root1);

assert(host2.getRootNode({ composed: true}) == document);

assert(host1.getRootNode() == document);

assert(host1.getRootNode({ composed: true}) == document);

[TAMURA, Kent]

LGTM1

--

TAMURA Kent
Software Engineer, Google

[Elliott Sprehn]

So if you're in the document composed: true always returns the document otherwise it returns the top of the subtree crossing shadow boundaries?

[Hayato Ito]

Yes. If we are *in the document* (in a traditional meaning, now it is called *is connected* ), getRootNode({ composed: true }) can return the document in O(1).

[Philip Jägenstedt]

Elliott, do you have any concerns with the options? I've taken a quick look at the spec and it looks sane to me, so LGTM2.

[Elliott Sprehn]

On Thu, Jul 21, 2016 at 11:54 PM, Philip Jägenstedt <foo...@chromium.org> wrote:

Elliott, do you have any concerns with the options? I've taken a quick look at the spec and it looks sane to me, so LGTM2.

Seems fine to me though I wish we would have added features in a separate step from doing a rename. :) Do we have some use cases or examples for the composed: true case? Is it just to save people from writing a loop? I guess that's okay since we can do it in O(1) for you. :)

[Hayato Ito]

Yeah, in a case of "nodeWhichIsNotConnected.getRootNode({composed: true})", this is just to save people from writing a loop.

I guess many frameworks have to write their own similar loop unless the platform provides it.

Note that adding optional "{composed: true}" parameter to this API does not contribute to decreasing the time complexity for this use case. It is not O(1) because we do not keep a *super-root* anywhere for a disconnected tree. :(

It takes O(N + M), where N is the *height* of "tree of trees" and M is "height of a (top) node tree".

[Chris Harrelson]

LGTM3

--
You received this message because you are subscribed to the Google Groups "blink-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to blink-dev+...@chromium.org.